This page is not created by, affiliated with, or supported by Slack Technologies, Inc.
Greetings from Finland and from Eficode! ๐
Good morning from the Berlin Watch Party!! ๐ฉ๐ช
Greetings from the Johannesburg Watch Party!
Hello from Melbourne. Iโd take a picture but itโs darkโฆ.
Bom dia from the Watch Party in sunny Lisbon! ๐ :flag-pt: โ๏ธ
Welcome @sam793, I'm glad we're endorsing the Eurovision concept of Europe!
Ha, @tom.baker and I are also in Munich, @irizinchenko. We should have started our own watch party ๐
Nick, for you it seems more night than morning. Good to see you here with us waking up as an US marines. :saluting_face:
Reminder: Get yourself in front of your browser because weโre kicking off the Summit now!
Good morning, everyone! Weโre starting in moments! Thank you all for being here โ we have a great two days!!!
good morning everyone!
hey folks, how is the video and audio stream looking and sounding on your end? ๐ or ๐
Greetings from New Zealand; Kia Ora - Excited to be here :kiwifruit:
Looking and sounding good at the Berlin Watch Party
I participated in this event last time in 2017. Somehow, everything is the same, but still, everything has changed ๐
amen to business outcomes and tech strategy being fully integrated - thats the holy grail ๐
^^ 100%! Business challenges and technological solutions. The two groups that need to be aligned and yet almost never are to this day
Reminder: Please update your Slack profile to include more information about yourself. Name, image, organization, title, pronounsโฆ whatever you feel comfortable sharing with this community to help us learn a little more about you. Hereโs an exampleโฆ https://devopsenterprise.slack.com/files/UATE4LJ94/F0704RVM1FC/slackprofile.png
Wow, @jeff.gallimore โ slick, posting the instructions on how to change your Slack handle! ๐
Ah, yes! Hereโs what my Slack handle looks like โ thank you for modifying yours!
How fun to be a speaker and get to talk and respond to the #C015DQFEGMT live!! :star-struck:
Credit goes to @genek and his (sometimes irritatingly) tireless efforts at creating this amazing Slack widget!
This time since we're live we're instructing speakers to get in here after their session ends โค๏ธ
Reminder: The Summit has a Code of Conduct because we want everyone to have an amazing time here at the event. Hereโs the summary: Listen well when someone else is sharing. Share well when you have something to say. Respect everyone at all timesโฆ and speak up if you see something or hear something that isnโt consistent with the environment we want for this community. If you have any issues, email <mailto:help@itrevolution.com|help@itrevolution.com> or direct message @jeff.gallimore. To see the entire Code of Conduct, you can read this post: https://devopsenterprise.slack.com/files/TASMB716H/FPW51DY5T
1If you @ me during my upcoming talk, we'll make an attempt to react live
Hello everyone ! My main question I hope to get before the end of the 2 days would indeed be : how do I get business leadership onboard ?
@alexis.gigleux in addition to the lessons from the talks, a couple of resources you mind find valuable: https://itrevolution.com/product/talking-business/ https://itrevolution.com/product/measuring-value/
(Some impressive timing on automated Slack posts from @jeff.gallimore โ ๐ )
1top right of the page with the video --> https://itrevolution.com/gather
Can people from the Watch Parties post pictures of the rooms? Thank you!!!
Yes, please! And please send to main channel, so all can see and envy! ๐
๐ Kicking us off this morning is https://etlsvirtual2024.sched.com/speaker/luca_ingianni.26lhbaot, Consultant, Trainer, Engineer and @tom.baker Robot Behavior Software Architect, Medtronic. presenting An Architect's Agile Journey: Insights and Lessons from Medtronic's Surgical Robotics Division
interesting enough - these medical devices have to run in both online and offline mode. makes the challenges even harder!
โWhat do software architects do?โ Visio? Powerpoint? ๐
I think that's a good distinction. What does the architect do and what should the architect do ๐
The famous SAFe Big Picture โ three places where software architects show up in value streamsโฆ but how do they interact with all the teams?
โWho should architects talk to? Who should they be interacting with?โ These are fantastic questions that need to inform how we โwire our organization.โ Fantastic! Need to help product and devs and quality!
โand how do we not become a bottleneck to their goals?โ These force us to a design area in the org wiring.
"feels good in theory, don't now how to breath this to life." thats the mindset I want all of my architecture partners to have, then we can figure out the life part together
I feel like there's a construction business idea in building "accidental ivory towers" - Brent the Builder? :thinking_face:
Would love to shadow a day-in-the-life of many different architectsโฆ seeing how the role is instantiated in each org by different people!
That would make a great blog series or self-publish book. Day in the life of 10 different architects.
Interesting @joachimsammer - I've seen vastly different footprints, interfaces, operating models and behaviours (the good, the bad and the ugly ๐ฌ )
I ve been in the isolated ivory tower before. Finding this really insightful
@phil.parker My personal impression is that at least at the EA and high-level solution architecture layers I observed different flavours (centralized, federated, enabling,...) but by large it was flavoured TOGAF ice-cream base in neat boxes. But I have spent a lot of time in financial services. ๐
You could accompany me, but half of the day would be watching me read documentation!
A big argument that always comes up if we talk about "rebuilding" architecture is that it will help us in the future to develop new features faster. Is there a good way to quantify that? Because weighting this future benefits against the business outcome of a specific feature that is "on the table" is always challenging. (I am a PM btw)
my mentor Chris Matts framed architecture as managing the future risk of the system - which I really like.. so you are looking further ahead about quality attributes based on product roadmaps and usage data / telemetry etc...
you create options and next best thing rather than design the cathedral that may never solve or deliver any business value..
It's interesting how that metaphor is broken. You don't build an architecture, an architecture enables you to build something. Perhaps that's part of the problem we see.
In the same way that we should think about technical debt as an analog of actual debt (interest rates and increasing costs if you're not making your payments) you can think about solid architecture as a financial investment, you may only be getting single digit percentage returns, but they compound.
metaphors are always floored, they serve a purpose but shouldn't be taken too seriously or people get lost in the metaphors and arguing about edge cases rather than the spirit of the game..
look at the impact of change over time.. if your lead time is getting slower over time your architecture isn't enabling you, it is slowing you down.. although that is also a function of technical debt too which can be a number of factors..
Grady Booch describes it best: > โAll architecture is design but not all design is architectureโ. > โarchitecture represents the significant design decisions that shape a system, where significant is measured by cost of change.โ This, to me, is a useful, more specific definition of how architecture is an investment in the future.
Hello from Lyon, France ๐ซ๐ท ! Beautiful day for a nice conference. ๐ฅณ
Creating boundaries in the system, to enable independence of action and isolation of effects (to prevent them from escaping, both from the modular boundary and into production)
Should we use TDD to help driving the architecture? Do u think defining the requirements upfront can help designing a better architecture based on user value instead of technology value?
I feel like this is something that the teams doing the work should help decide ( through techniques such as DDD). If this is something they haven't done before I think an architect with experience can really help guide the conversations in the right direction. Personally I would also say that creating clear boundaries is a good aspiration, but thinking you're not going to have dependencies or gray areas between boundaries is not realistic and will likely fail. Instead aspire for that isolation but understand dependencies and embrace the techniques (through things like TT) to better manage those dependencies. This is again where architects I think can provide guide value. This is the socia-technical challenge. I think lol ๐
Help manage and plan for managing the cost of change โ love this.
I'm always wary of roles describing themselves as "a bridge". My snarky comment would be "many people who describe their role as a bridge, end up acting as a moat"...
I think there is great value in communication, enabling collaboration, but it should really be bringing parties together into a shared working space, rather than being the link between them?
There are some words here which seem to indicate a divide between tech and "the business" - eg "the two Sides". Seems to go against the point below. At Atom Bank I have banned the term "the business" as it subconsciously creates that divide.
@andy.sturrock I've done the same in the past - although in my latest mission we are very much recognising that there are people outside of technology who have specific commercial/operational responsibilities (collectively "the business"). However we've been working really hard to make sure there is shared ownership on the value delivered from technology/digital/data initiatives.
Oh for sure. We have a Commercial team, a Finance team, an Ops (customer service) team etc. I just don't like calling anyone outside Tech "the business". We're a digital bank, so Tech is at least as much part of the business as any other team.
Good points. Yet the divide doesn't exist for no reason. These people live in different worlds, use different vocabulary. Just telling them to be friends doesn't work, somebody needs to make the introductions, do the translating, watch for misunderstandings. If they make themselves superfluous over time, that's awesome, but whether we like it or not the divide is real.
And this is why an architect stuck in the middle isn't helping: then they become the moat you quite aptly mentioned.
Thanks for replying. Yes if there is a divide there's no point pretending it's not there. So as per @phil.parkerโs point the role is to bring together rather than "interface". Language is important which is why I don't like "the business" as it reinforces the divide. We've had real success bringing the teams together at Atom over the last couple of years where we now have truly multi-disciplinary teams (so people from Commercial, Ops, Tech - not just multi-discipline within Tech eg BA/Dev/test) all working together to create new products and features for our customers. That's way more fun than "the business" telling Tech what to do ๐ .
I see your point about "the business". Ironically, I used it as a shorthand for, I don't know, "counterparts" to the technical teams, but indeed what we called "business" isn't the business or users (i.e. doctors) either, they're sitll one or two levels of indirection away. I suppose it was shorthand for "people whose perspective differs meaningfully"
That's why I just say to use the name the other teams call themselves. So if it's the Commercial team (the people who decide what our interest rates should be etc) then we just call them Commercial. It's all a bit meaningless otherwise - eg our People Experience (aka HR) team call the Tech team "the business" so the term really is ambiguous. Anyway, it's just one of my pet hates along with calling people "resources" ๐ . Thanks for the talk - it was really good.
> Anyway, it's just one of my pet hates along with calling people "resources" Heh, I hear you
What Iโve learned is that EVERYTHING is about about communication and information flows, whether manufacturing or software. This is great. (And thus the need to wire our organizations well, which in our context, is all about architecture.). Nice work.
Shouldn't developers be close to product management though? The more intermediaries, the more communication channels, the more confusion and miscommunication.
Would it not be better to have Developers and PM/Business talking to each other directly rather than funneling that communication through Architects ?
I think a key point is scale of the system relative to the number of teams involved..
Architect as facilitators and translators of communication between rolesโฆ
Should architects be hands on? At least prototyping? How can a hands off architect understand the impact of its decisions in development productivity?
Interesting @joao.franco. So an architect at least performs a POC before sending over to Dev?
IMHO should at least experiment / spike in the code base, bc it will make a more informed decision that has direct impact in the development productivity
A classic example is when normalising databases, it can have a massive impact on read queries performance but from a theoretical pov it is the perfect solution, also from a development perspective when thinking about testing it is very often much more complex because the setup requires creating much more entities
Joao, this is what Tom has ended up doing, and both him and the developers have found it useful
@genek I โpervertedโ the 80:20 rule. Software engineering is 80% social and only 20% technical. This great talk illustrates this perfectly. Thanks @luca @tom.baker
Boxes and arrow โ funnily enough, one of the best way to portray connections. Nice!
facilitating shared understanding through conversation, coaching, and boxes and arrows - enabling teams not making technical decisions on behalf of others.. or you lose shared ownership / understanding (someone elses decision)
With all this talk of boxes and arrows it would seem remiss not no point out the success we've had with https://c4model.com/
The main thing is connecting, facilitating, and asking the questions, which will help bring everyone to a shared understanding and use the wisdom of the group
Boxes and arrows diagrams can be a super helpful tool in ensuring that everybody is on the literal same page and not making assumptions that will trip you up later. They're a helpful tool to help with the comms and connections that a modern tech architect needs to do.
โarchitects should use the same tools as the software developersโ ๐ก ๐ ๐ ๐
100% Agree, it's not easy to get the developers and business speak the same language. Architects ๐
Developers can talk to the business, itโs self importance to say they canโt.
we (my former role) tried not to use the term business as anyone working in a company is part of they business - moving to product and customer orientation focuses on needs and opportunities rather than some group of people from another planet ๐ they all need to meet in the middle - quite often product management doesn't exist and architects try to fill that gap
Aren't developers part of the business? (I mean, as a consultant I know it's a bit of a different story, but in other cases ...)
Part of a team, yes, if having to filter conversations via architecture then other problems.
Oh of course they can, and they should. I didn't mean to imply that at all. However it's my experience that they often find it difficult to understand one another well
My experience is the opposite once that relationship is formed. Less on roles more on enabling that direct conversation should be the focus.
I wonder if speakers have thoughts on the architects key role to be that of broadcasting understanding, one by which that understanding can be communicated to different audiences (regulation, audit, teams etc) and in different forms dependent on need?
Yes, I think that's a crucial part of the role: translate between different points of view, languages and indeed world views
Yeah such that it can be queried, adapted, fixed, improved then you might create that sense of co-ownership
amazing talk @luca & @tom.baker โค๏ธ
โก Next, here to present Pfizerโs Future of Development is @jamie283, Director, DevOps Lead from Pfizer, and and @laura507, CTO at DX. They will talk about how developer experience became so critical at Pfizer, and how they helped their organization win, eventually helping thousands of developers across the organization.
Thank you, @jamie283 and @laura507! I remember so well talking with Brady, a UPS driver, right after he delivered the first COVID vaccines to a small Oregon town in Jan 2021 โ I remember tears welling up, hearing about how momentous and important that was, the first step to bringing normalcy to society. ๐๐
โit eventually became more difficult to maintain technical excellence.โ (Ouch. Because we were too mired in trying to figure out how to write YAML config files and such, is what comes to mind.)
YAML the modern equivalent of raising a request for an on-prem server (painful)
Retaining key talent is key to continuing success
Right, this topic is also extremely broad, encompassing even ESOP measures
Right Nick, looking at last 2023 Gallup engagement report and the last 2023 burnout report from Infinite Potential, this should be one of the highest CXOs priority
burnout yes. I would not say stock options would motivate great work just sticking around
โwe used to outsource our tech strategy to external vendors. this turned out to be not smart. this needs to be led by our most senior engineersโ ๐
@jamie283, my experience of large companies such as yours, is they tend to ask for such a large scope of technical knowledge in job descriptions that it becomes almost impossible that anyone would really have all that knowledge. What's your view as an insider?
I will admit, some of our Job descriptions are terrible. I think part of this is a result of trying to ensure that the role is deemed senior enough internally to justify the level requested. Roles often go to an external group to be graded. My advice here would be to try and pick out what you consider to be the key needs for the role and apply if you meet them. I will take this as validation/feedback though to think about!
Principal Engineers!!!! โsome of our most trusted and respected peopleโ
Nice: On the one hand, we discourage โfungibility of contractorsโ, but also we recognize that at our scale, we have people onboarding and joining and switching teams all the time.
Making "the right thing easy" is the "paved paths" concept in platform engineering.
In addition to writing, would training AI as a companion be useful?
We have teams experimenting with AI for sure. Chat based interfaces for documentation is one such experiment. I don't think AI (or at least I have not seen it) is ready to replace a great technical writer or a passionate engineer who wants to communicate. It can certainly help though.
@jamie283 I'm interested in hearing more about what you mean by "discourage fungibility of contractors".
creating internal mobility and learning of employees seems a lot more useful to building capability in a large complex org.. if you are leaving it up to contractors you are you have a lot of extra risks to manage..
1I am not sure this is true of all large organisations, but with Pfizer we do have a large contractor base. It has been consistently growing since I have been here. What I don't like to see is a rotating door of contract resources, if there is long term consistent demand but we cannot convert that spend into employee headcount. It is better to retain the contractors and plan their moves around the organisation vs new bodies consistently.
1very interesting @jamie283 - what is preventing converting to perm / hiring - is it s project funding (OpvsCapEx) / thinking challenge or something else (super curious) I can understand a need to ramp up and down some people over time but balancing .. in terms of capacity you cannot simply onboard productive people and ramp back down after 6 months..
Yes it is to do with flexibility & the types of budget used. Also there is an assumption that we need to flex more than we have demonstrated in the past. For the past few years we have been hiring more technical resource as employees. This trend is continuing.
The two talks together triggers the question. Architects vs Principal Engineers. What are the differences ? Do we need both ? Is there overlap or are they totally different ?
this would be a fantastic networking/breakout topic. At my company, Target $110B USD company with over 7k technologists - we have 6 total architects and about 100 Principal Engineers. I see a stark difference, but would be interested in other views
This is a great question! I find that Principal Engineers of the 2020s are doing things that Architects have stopped doing or donโt have the skills to do. Using techniques and thinking from DDD, Team Topologies and even Wardley Mapping. But at the same time have the credibility to influence and coach teams on their ways of working and technical architectural decisions. Not dishing architects here but in my experience the โPrincipal Engineerโ role is closer to what is needed than the โArchitectโ role as we have understood it for years. Ultimately people can move between the roles
For me the "struggle" to define what an Architect should be doing comes from the fact that Architecture design etc. tend to be cyclical work. It's a hat more than a role.
I think it's important to distinguish the title from the purpose, because I suspect we all can agree on certain functions we feel need to be taken care of, but where to draw the boundaries and what to call the different roles may vary strongly even within a single org
Indeed at Medtronic for isntance they need a very technical engineer role -- it would probably make sense to call it principal engineer instead
I usually see Principal Engineers doing quite a lot of architecture work.
Also, Architects with a solid foundation in software engineering are the best. You should be able to at least tinker with the systems you're integrating to prototype things and fully understand how they can work together. Most of the Architects I know were software engineers first, and the best ones still get their hands dirty at least a couple of times per month.
When you get to a certain level of scope, though, it becomes too difficult to work across many products and still take up any story-level implementation work. You find that you never want to be the bottleneck or have anyone waiting on work that you're doing.
It's a grey line at Red Hat at least. I think most people with the Architect role here are graded in the HR system as "Principal Software Engineer" or "Senior Principal Software Engineer". It's not a totally separate career track.
We also generally let people write their own job titles - as long as their manager agrees then it goes into the org charts, and you make your own business cards, etc. Someone might choose to change their job title to Architect because they feel like they're not doing much coding, or because they like how it sounds better. Others are allergic to the term Architect because they think that those are people who don't do any real work, so they prefer SPSE. So to me, it's more of a role designation.
Exploring how to go WAY BEYOND mere advocacy. Nice!
What tips would you give those looking to transform how they work to get real buy in and support from execs to make it a reality?
find business problems / opportunities that are worth solving (apply commercial lens) and focus on that - avoid talking about methods / frameworks etc or you will quickly lose people
@vladyslav.ukis Chris Matts calls this the zeroth constraint - similar idea, prove some value, establish trust by solving a problem rather than selling a solution https://theitriskmanager.com/2017/02/05/the-zeroth-constraint/
Love these โ in other words, โall the things that developers donโt really want to worry about.โ (speaking from personal experience!)
Would โidle timeโ be more accurate then โtime lossโ, it sounds to me that idle time is easier to measure then time loss because they can be interpreted differently. Idle time can be for example how much time a dev is waiting for pipeline feedback.
Time loss is idle time, time lost due to friction, rerunning a flaky test, rerunning a CI build, wasted time in a meeting, etc.
I get it, I was just wondering if in terms of measuring, the way u measure each one and how u act on it to improve are different? One can have a more technical solution and another a more organisational/cultural solution
Yeah, the time loss metric is a driver, which is sort of a โbig pictureโ metric. Along with it, Pfizer is capturing specific signals about where that time loss might come from (CI/CD, test instability etc)
@laura507, do you think there's a way to incorporate DevEx into the onboarding process, 'shifting left' on DevEx almost?
Yes, onboarding definitelyโฆ in some ways I think that if you have an excellent onboarding process (think lots of documentation, examples, service catalog, golden paths everywhere) to get a developer up and running quickly, those will also improve the developer experience of every other developer in the company.
By in the recruiting process, do you mean highlighting an excellent developer experience as a way to attract highly talented folks?
@laura507, thanks for your replies. What I meant is organizing the recruitment process in a way that allows the developer to showcase their 'talents' and their fit for the particular role or project in question.
Yeah, there are ways that work better than others, of course depending on different companies. Iโve reworked engineering sourcing/recruitment/interviewing for a handful of companies. I found that candidates rate the interview experience more positively once they have a chance to ask their own questions, and talk about their technical experience in detail, before being asked to do a coding interview. Usually those two things are enough to see if itโs a good mutual fit before going into the more time-intensive parts of the interview process.
@laura507, thanks, I've got to admit that's the least I'd expect, but after my experience of various if not many interviews, is that you can ask all the questions you like but you don't necessarily get the most convincing answers. I wondered if there might be some kind of pre-interview process to focus the interview towards aspects that the developer has questions about with the job description so the 'odd' slide could be shown in the interview or perish the thought, even a small demo could be arranged to give the idea of what they might be working with. I know intellectual property questions might come into play but it would be good for both the developer and project team to get a measure of who they are engaging with.
Back when the market was more competitive for talent, Iโve seen many companies take that approach and focus more on selling the role before getting into skills demonstrations.
Well, I'm glad at least that your work is at least putting the question of DevEx on the table. Keep it up!
Greetings to Phizer, great presentation and the content. Question: you measure the success by assessing Project and Program metrics. Should not that be Products instead if you decided to inhouse technology and treat it as a key differenciator?
Products sit at a different level. The program is large and touches the whole org, projects are very targeted initiatives that touch multiple products.
Yes, let me rephrase the question: how do you measure success of the products? Is it part of the transformation metrics?
the bus problem is a good place to start >> how many people have to leave before the team cannot function / deliver
I had a colleague who refused to use the bus problem because they knew someone that actually did get hit by a bus. So they renamed it to the circus problem. How many of your people need to run off to join the circus before your team is affected ๐
@kevin.ashton I totally agree and usually reframe it as how many people hop onto a bus and leave.. rather than "hit by a bus" which is not great language - def agree to your point
attrition risk in this context is how many developers are at risk of voluntarily leaving the company because of low satisfaction
not how much risk attrition would bring to your companyโฆ. but that is an interesting one ๐
How are you measuring developer satisfactionโฆ โvoice of the developerโ can so often be overlooked, really important to listen and hear what drives the attrition and then act on that. Really interested to hear how you measure and tack this
They measure it through DXโs survey tool, which is โvoice of developerโ and captures satisfaction, but also has questions about workflows (just like DORA is survey based but asks questions about deployment frequency etc)
@laura507 does that cover the whole team (Cross-functional - including product / design etc) - how are the cross-functional teams managed and made visible into the DX tool? quite often tools dont have a real cross-functional team and structure that is visible and consumable across tools.
@chriscombe definitely a conversation Iโd want to have with you and your team ๐ DX targets developers right now, so most orgs donโt include product/design because the questions donโt apply to them.
We use a survey tool tooโฆ in fact a number of surveys, one corporate survey twice a year, a monthly survey at an Engineering domain level and a team health for each team on Slackโฆ in addition we hold listening sessions and skip level meetingsโฆ it sounds a lot but itโs important to us to hear what our value creators are feeling/experiencing
It would be interesting to hear if Pfizer see the transformation as a journey or destination? At a sponsor/VP/C level that would be interesting to hear
I would love to hear a bit on how Pfizer defines the Principal Engineering role.
We currently have a few posted roles, this might help frame the expectations of the role. https://www.pfizer.com/uk/about/careers/job/4908955
Terrific presentation from Pfizer, loved the DevEx metrics and what a journey you are on
Great presentation! Do you have an example of a signal from DX that made you change an approach or direction in your program? @jamie283 @laura507
Documentation contributions would be a good example of this, also knowledge silos ( โhow often does it take you 10 minutes or more to get an answer about (code|system|etc)โ) The program emphasised knowledge and training for the first push of work and both of those numbers went up over the last 1-2 quarters
๐ Please welcome, @ojacques2,Sr DevOps and Cloud Architect, AWS, @claude.ampigny, Sr Customer Delivery Architect, AWS and @clementsabater, Delivery Manager, EDF. They will provide a behind-the-scenes view of the digitalization program of one such large industrial organization, and how they adopted innovative strategies such as the "You Build It, You Run It" model, "Team Topologies" for organization, and Amazon's famous operational mechanisms.
https://devopsenterprise.slack.com/archives/C06UL60MN1Y/p1713949990957369
Hello, @ojacques2 @claude.ampigny Clรฉment Sabater on how they are LITERALLY keeping the lights on! ๐ ๐๐
Doesnโt France have a large amount of nuclear :radioactive_sign: energy production?
The changing energy mix โ and Iโve heard @adrian.cockcroft talk about the insane power needs of modern LLMs, and how that is putting unprecedented pressure on data center sustainability.
Are you really projecting a reduction in energy usage ?? Everything I've read and understand about where things are going led me to believe that the worlds total energy needs will be going up exponentially. Would like to understand why France is different.
probably nuclear and renewable rather than reliance on fossil fuels (total guess)
That is a good question! I have heard that Generative AI will be one of the big drivers of this increased use so the focus must be on energy efficiency.
Great question. It is indeed the projection. The global consumption is reducing. The intention is that in France, there are massive investments in building / housing efficiency (up to be positive). The grid is also being decentralized with self-production (solar). Full report here, coming from government (end of 2023). Other projections are saying differently. The important piece of information is that the mix is drastically changing, so MUST IT. https://www.ecologie.gouv.fr/sites/default/files/23242_Strategie-energie-climat.pdf
@ojacques2 @claude.ampigny and Clรฉment I think this is so exciting โ Iโve interacted with people in the critical infrastructure community and some of these tactics just seemed unthinkable to them. I canโt wait to share your story with them!
There are nuances of course, and not everything is in scope. Yet, to run an industry of this size, which so diverse, there are plenty of opportunities. Opened for discussion!
The Equal Experts Berlin watch party are enjoying the morning talks ๐
Dr Werner Vogel with the Half-life t-shirt looking rather cool!
In critical infrastructure, you have not only the typical IT silos, but you also have the IT/OT divide, which we have an amazing talk on tomorrow โ another amazing first.
@ojacques2 @claude.ampigny @clementsabater What technologies did you use to create the consumable components and to bring them to the S.A. teams?
If S.A. stands for Stream Aligned: it's all about the golden paths. A golden path for Single Page Application (SPA) type, a golden path for private web app, a golden path for feeding enterprise data lake. Another dimension are golden paths for the experience from developer environment, to build, to CI/CD, to operations, for each of the main technologies: serverless/lambda, containers.
S.A. did indeed stand for stream-aligned. Thanks for the answer. I might have missed it a little in the presentation but with golden path, you mean all the tools / components required to develop, build, integrate and deploy each of these types of applications, right?
โas part of our Press Release preparation, we create list of questions that all our stakeholders and customers might haveโฆ an incredible challenge that takes a lot of work, and forces us to understand who our customer actually is.โ
This also relates to the "https://www.youtube.com/watch?v=L227qFemjqI" which happens when we share the PR/FAQ with stakeholders, and opening for comments. We want the FAQs to cover everything which can be asked.
Love the idea of the PRFAQ approach but I suspect it needs the right culture and discipline for it to work. Has it worked for you? What are the gotchas?
I love this approach. I was working one this morning. You need to have discipline in writing and the ability to partner with a few strategic people. It's not an operational thing, but very effective.
First time it was surprising and the experience of aws proserve was a must. But two iterations later, we saw the value especially to correct our misalignment. I confirm itโs more a strategic approach .
(A fantastic example of Slowification: pushing activities into Planning and Practice!)
โbuilding apps that will last for 20-30 years. yes, we want to optimize for devs, but also for ops.โ ๐ฅ
I guess even more so when those humans based in ops could experience a 100% turnover of staff over that timeline!
I like the aspiration but is this really sustainable? Has the change in the tech environment slowed down? Will apps delivered in 2024 not be legacy slowing us down in 2044? And massive tech debt? Isnโt what we want to optimise for changeability and evolvability?
I recall a Gartner expression that the majority of code cost is post implementation, especially if something goes wrong and lots of unhappy customers ring your contact centre. I also liked the expression early to insulate against the cost of change. This rings true when you couple with third parties to provide services and you need to change over time.
> I like the aspiration but is this really sustainable? Has the change in the tech environment slowed down? Will apps delivered in 2024 not be legacy slowing us down in 2044? And massive tech debt? > > Isnโt what we want to optimise for changeability and evolvability? Any new code that we write is the new legacy anyhow. First thing is to move a project based approach to a product based one. It's a departure for this program. Regarding tech debt : the intent is for the product team to handle it, helped by automation for example to upgrade run times, and dependencies. Managed services also helps to move tech debt to someone else (AWS in that case).
@ojacques2 ๐ฏ offloading to someone else e.g. AWS, definitely extends the half life of the code we write as you should be largely able to rely on the contracts/interfaces being respected over time. Also, runtimes, ops, security managed for you etc.
โweโre launching one new product per monthโ. (!!!)
Slowification!!! such an important thing that I never had a word for until the Wiring the Winning Organization. Someone posed the question earlier about walking away with how you get business outcomes aligned. It takes regular Slowification, deep thinking time and over communication. I love the example these guys are sharing here!
So cool โ hiring goal: 350 cloud native devs to support critical infrastructure for power generation!
Do the need them in those places or does #FullyRemote work align with the org?
Seriously, @clementsabater - we at EPAM can help with 350 developers. Ping me and let's connect/discuss.
We are just starting and we want to create a strong team spirit and we preferred to live together 2 or 3 days by week for the moment.
It is our hypothesis. We want to iterate - do you see the same?
It seems to show that platform engineering is applicable to skills only. Is this what it shows?
We did not want to decompose the equation but in fact: platform engrineering = Skills ^ Builder Experience. The point we wanted to make was that platform could help reduce the skills level.
The challenge is not attracting top talent, itโs developing top talent. https://youtu.be/NAazenCRQSY?si=_vhqdMUtgpL4AEQZ
Agree to some extent, sometimes I find challenging developing the atitude that leads to wanting to learn with which u can easily develop talent
Attitude, and character. For example, โข Is the person trying to be the smartest person in the room or to make the room smarter? โข how curious are they? โข Do they ask good questions? โข How do they treat failure? Personal and organizational
Right! Going from where we are starting, remember that the context is that all build activity is outsourced, and will mostly continue to be.
Ping me for 350 cloud developers - we can help with that!
Iโd love to chat about that DevEx book and what need to be in it. I have a pipeline of ideas and would love to validate them
Everyone seems to be hiring right now. Demand is strong for the right skills. The media hype on layoffs is misguiding.
What is the power defined as? And why is skills and ops model (and not architecture) in the power?
The power it our acceleration factor. And improving the Builder Experience (aka DevEx) as actually a very high impact on our velocity. Architecture is not part of the power because the Builder Experience (aka DevEx) as no impact on it.
One of the questions I would love to know more from the next two days is how are teams budgeting for their approach? A lot of CFOs I've worked with will drive conversations around "what features are we going to get and how long is it going to take to delivery them based on a set investment". I feel like that's at the root of driving teams to work in the wrong way. Pushing back usually elicits the response of "how can you run a business if you don't know what you are going to deliver". Would love to know others experiences, whether they've encountered the same and how they've approached these conversations.
moving to a capacity funding model can be helpful - Mik's book on project to product is a good place to start as well - it is a deep topic but happy to chat if you want to share stories / thinking
At Atom Bank we have basically a fixed cost for Technology that we (exec team and board) decide on a yearly basis. So we still get asked "when will the new savings product be ready for launch" (which is totally reasonable) but it's much more a capacity funding model as per Chris above than project based funding. Then it's my job to ensure we maximise the output/outcomes (new features/products/etc) we get from the input cost.
Transformed by Marty Cagan discusses objections by different groups of stakeholders/leaders as well. Shifting to a capacity based model in combination with lean portfolio management and a product-centric/discovery-driven approach can work. Variations of the question 'what is the cost of delivering the wrong things' can lead to interesting conversations.
another hack is to crate transparency on just how made up / useless time booking on activities can be (in extreme cases) I remember standing in a room with the CFO for our tech area (former employer) and telling them I had to book time against 10 different projects and I wasn't working on anything of them and the number of people with the same situation basically created a bit of a mind blown type moment.. accountants / financial people love precision and accuracy, I leveraged that to say we have precise data but none of it is accurate as people dont book time (internally) for any meaningful reason other than because they are told too
if you use that to drive your internal billing and allocation that can drastically impact the defensibility of the finance department e.g. why am I being charged internally for project xyz ..
capitalisation of SW assets is another area to look for opportunities too in terms of depreciation, amortisation etc.. lots of reg levels to play with
Thankfully we have a great CFO who says it's Finance's job to worry about Capex/Opex etc and that we should focus on building great tech for our customers.
I can also connect you with the former CFO I worked with if you want to connect your CFO
There is a useful case study on Capex/Opex automation in 'Product Operations' to see how a public US company operationalized that.
โน๏ธ I see a lot of people stuck at the staircase. Please press G to become a ghost and pass the bottleneck.
@alexb is there a way to make Ghost mode the default, so one can pass thru bodies without needing to press the G key?
it isnโt possible. Itโs a โworkaroundโ in GT bcz when youโre a ghost you canโt interact with the other people.
Reminder: The talks are starting again in 5 minutes. Start navigating your way back to your browser. https://devopsenterprise.slack.com/files/UATE4LJ94/F06V49GSELX/timer.png
๐ Up next is @mili.orucevic, Chief Software Quality Engineer, and Alin Iacob โ Cloud Architect, from Visma Software International, here to present: How We Are Wiring Winning Organizations In Visma. They'll be talking about their grand journey that theyโve been on since 2015 to enable greatness in their organization, and better serve their customers.
1Ten years can go by fast!!! โin 2015, we had lots of silosโ. So familiar!
most definitely. Its so important to break down org structure and silo'd incentives and align on common outcomes, horizontal thinking, and leaning into the diversity of skills/experiences that each person/function bring
I love these depictions of Dev vs. Ops โ what was amazing working on โWiring the Winning Organizationโ was that the same dynamics exist in almost every organization in every domain in every phase of value creation. ๐ In pharmaceutical development, you see the same dynamic between chemists and molecular biologists and supply chain. In โTeam of Teams,โ we saw the same dynamic between intelligence agencies and military services. cc @laura507 @tom.baker
And @lucas.rettig talked about this two years ago in supply chain, and the WAM vs. WOM. ๐
I'm about to attend our WAMWOM (weekly action meeting vs weekly operations meeting - now 1 meeting) in 2 hours!
A first from @mili.orucevic and Alin โ switching off screen shares!
Interesting! Slowification help especially with the non-functional requirements!
More surprising and wonderful Slowification practices โ but also building connections and relationships. Very nice.
So interesting on the prominence of architecture groups today!
I am curious - who is using continuous discovery techniques - e.g., opportunity solution trees, assumption mapping?
check out learn wardley mapping from Ben Mosier - he's got great stuff ill share a link https://learnwardleymapping.com/
What comes to mind, is are you including in your discovery techniques Outcome Hypotheses to validate and learn from your โdiscoveriesโ and assumptions?
Hi @mili.orucevic. Has the VATP Testing Capability Assessment been published anywhere?
Hi @steven.knopf - no, all of our assessments are internal and only available to the Visma companies for the moment. But a great question and something for us to consider, open sourcing it ๐ก
The London Watch Party in the centre of the city. โค๏ธ
Wow, check out that trend on the bottom. Amazing, @mili.orucevic!
for deployment frequency that's shown here, we're aggregating data across 3-4 years, from around 140 teams
@mili.orucevic When Alin says, โdiscussion with the Boardโ โ which board is that? Board of Directors?
Yes, that is the board of directors of the different companies we have in Visma.
โmore freq deployment ---> FASTER ARR GROWTHโ. (!!!!) ๐คฏ๐คฏ๐คฏ๐คฏ๐คฏ
I'm extremely happy to see empirical data on this, because people frequently push back on the notion, and more arguments are very welcome.
I have seen that time and time again particularly for internal products / applications.. e.g. we can only update the general ledger once a quarter because we cannot change too quickly..
@luca I think itโs partly because with old methods higher speed tended to mean less control and less safety which made businesses wary
@chris.leeworthy507 And it did in fact mean higher risk I think, because that slow cadence was really the maximum the organisation could safely achieve with the practices that were in place. Now the practices (and hence the risk profile) have changed, but the sentiments haven't kept up
More frequent changes means โข Less risk (only 1-2 changes at the time, instead of 500 changes ๐คฏ) โข More feedback from users / customers --> Easier to adjust and happier customers โข Low change friction - means your delivery pipeline is most likely highly automated
If you ever write those many aspects up or present them, I'd love to get a heads up, @mili.orucevic
@dave.watson are you referring to the people supporting the teams, with the different programs?
Yes, of course ๐ For some teams it might be overwhelming in the beginning, as Alin mentioned, things that are f.ex. mentioned, "why should we do all of this"/"do we need to do everything?" - but with so many learning and successes from other teams, one is usually convinced - but keep in mind it is not mandated to go all the way to the top of the mountain - some teams can not achieve that either, but they can get a lot of value just doing the security and architecture & technology program.
I'd like to propose an alternative title for "Wiring the Winning Organization" based on personal experience, this talk, and others we're hearing to "How to Align and Get Sh*t Done Better and Faster"
isn't that a different version of better value, sooner, safer and happier ๐
@kevin.ashton โข https://www.visma.com/blog/the-visma-way-of-working/ โข https://www.linkedin.com/pulse/developing-delivering-operating-software-retain-talent-lystad โข https://www.youtube.com/watch?v=neT2P81JKXM&t=1137s โข https://space.visma.com/channels/103055360/GlobalMessageBoard/article/77895602 โข https://www.visma.com/blog/vcdm-101-how-we-make-your-business-software-future-proof/
โจ Let's welcome @barbara.arnst, Transformation Leader - Organization Designer at Telenet, one of Belgium's leading telecommunication providers. She's here to present, ReWiring the TelCo Operating Model: Learnings from Telenet.
That was great. With such things we have to remain mindful of Goodhart's law. "When a measure becomes a target, it ceases to be a good measure"
โฆah, the Spotify model trapโฆ. โBoy, we jumped on that ship, brought in consultants, built blueprints, etcโฆโ. Shifted people into chapter, tribes, etc. โtwo years later, we took stock of where we were.โ THIS IS AMAZINGโฆ. can you guess what comes next?
โข more dependencies โข frustrations โข simple changes required coordination with 30+ teams
I was in a meeting to discuss a project plan to plug in a network cable the other day. I feel this @barbara.arnst
And it's a shame, because agile is a great way to work and deliver.
very true but people conflate doing agile (the processes) as the goal vs learning, feedback loops, customer / business value etc.. it is very alluring to buy into a "model" because look so many docs and nice words.. but orgs need to learn for themselves that works and is unique to their context.. (sorry soap box topic ๐ )
Easy to say, difficult to actually do. Sometimes I make an exercise in trainings, asking people to describe how they provide value. Invariably I get back a list of activities ("I reboot servers") instead of a list of goals ("I ensure the system is avaliable") That way of thinking really needs to be exercised before it solidifoes in minds
Completely agreeโฆ you shouldnโt do it if you donโt know how it will add value to the customer The Power of โWhyโ
"customer centricity > cost" so good. do the right then, then do it better, not the other way around
Iโve seen firsthand now the impact of enterprise-wide reshuffling people at massive scale to โimplement the Spotify modelโ, often triggered by external consultants. Iโm in awe of the chaos and destabilizing effects of it. This will surely be one of the themes in Vegas later thisyear.
I like the question: Where is our energy going? Is it to customer results? Forgot the alternative she mentioned, probably due to PTSD
(This programming focus was inspired by seeing this happen 3 times last year, and then triggered by seeing Barbaraโs presentation last yearโฆ)
โCIO function no longer exists โ that role has been distributed into the business unitsโ
this is the type of partnership and federation that is needed to bring IT and Business (or product) together and co-own the outcomes
this is how to remove the us and them language and build true partnership
Even Henrik Kriberg said the Spotify model is NOT a model: https://blog.crisp.se/2015/06/07/henrikkniberg/no-i-didnt-invent-the-spotify-model
This really looks like what you would need to take away from implementing Agile at an Enterprise scale!
I wonder what about human incentives in this mind-set shift? It is not easy to push people from Managing to Owning.
czeลฤ Marcin! it's all about the culture, set of values and beliefs the team works by, and the leadership profile that drives the company, ... once there culture, strategy, and mindset align, people start to choose the better solution (pull) over needing to be told what to do next (push)... and along the way, we can, of course, show that technology and operating model changes make us efficient and effective...
I think the shame around the Spotify model is in part related to how we naturally reduce a problem down to something - reducing the problem to it being "the Spotify model" in a way unfairly discredits the ideas shared by Spotify in those videos. Potentially more of the problem is related to the context and execution and not the model/ideas itself. Maybe I'm a laggard but if I detract the actual execution, the ideas on how humans interact and what environments might be productive - alignment and autonomy still really resonate with me and environments requiring creative brain work Guess I still relate it all to Dan Pinks Autonomy, Mastery and Purpose
For sure โ itโs not the Spotify modelโฆ itโs how (often consultants) sell an idea, and then execute it. In fact, Iโve started thinking of this as โFinding ways to fix problems caused by external consultants coming in and wrecking my entire organizationโ
it resonates with me @genek from my time at a certain large Swiss bank
the Spotify nomenclature (tribes etc) is so pervasive with my corporate clients now, frequently without a clear understanding of its background
Resonates! Spotify ideas are solid, but โblindlyโ copying without enough context and deep understanding will result in disappointment. I am so glad our company went thru that learning curve - without the pain we wouldnโt be where we are now. Consultants can play a tricky role in this - they are often pushed for quick/shiny result stories - resulting in the โold wine new bottleโฆโ.
i think iโve heard or read that somewhere beforeโฆ :thinking_face: ๐ #C058GHCP3FS
โmore than just being visionary on business side โ but also understand constraints, enable, etcโฆโ
โOur CEO is a believer โ it brings tech and business togetherโ
โslowification and transformation now become part of BAUโ
No end to a transformation - it's a journey to continuous improvement!
Totally agree, as a "result" of the transformation you should have an organization that is able to adjust to the ever changing needs of the industry. Processes and technologies and constant imporvement
If you get a consultant that promises to complete a transformation in 18 months, run...
โbudgeting time for operating model improvementsโ ๐
Trying to articulate issues a previous company were having with the spotify model, I found this which has some excellent points and reflections in it (written by an ex-spotify person) https://www.jeremiahlee.com/posts/failed-squad-goals/
Quote from their CEO: โโI firmly believe we are in a strong position now, not only in terms of clarity of where we want to go to, but also in how we will get there. Our Operating Model empowers our Tribe leads with all the means to be successful and deliver their ambitious customer promises. They can also fluidly adapt where needed, ensuring sustained success of the model.โ John Porter, CEO Telenet
There are several pervasive terms that I find quite unfortunate because of their connotations, "Transformation" is one of them (because it implies an achievable end state). Another one, BTW, is "Value Stream", because it implicitly seems to reject feedback (since streams are directed strictly downstream)
And then I need to expend a lot of effort to try to uncover and correct those unspoken implications
:flag-au: Next up, @sam793, Founder & CEO, TeamForm, here to present: A Benchmark Of How Hundreds of Organizations Are Wired and Mis-wired. He's here to discuss common patterns and configurations, and how nearly all organizations are missing some piece of critical โwiring.โ He'll talk about these patterns, and what leaders can do about it.
@barbara.arnst - that was really interesting. What sort of roles and mindsets work well in your enterprise tribes?
OMG. โLetโs start with a scenario.. Imagine youโre in an organisation where light switches are wired to the fire alarm.โ ๐
How often does this come down to: "The lamp on my desk doesn't trigger the fire alarm, so why should I fix it?"
โskilled security engineer billing time to 100+ projects to ensure she keeps fundingโ
I give gratitude every one of these events that booking time to project codes is not something that is part of my company's DNA anymore
โ80% of orgs have cross-functional teams: it is a fundamental way we create value.โ And yet, the only way theyโre recorded is in a spreadsheet.
yesโฆ because the excel file is serving as a database for other processes and applications
๐ Would love your participation needed: If i can ask you to jump into this link or scan this QR code https://www.menti.com/alxp9w1yrmf1
The work of the well-meaning Excel craftsman. Might indicate a cause or contributor to poor-wiring.
nothing can go wrong with the vba code pulling from those db2 sources in the back
This was great โ this picture was from the ETLS Forum here in Portland. It was such a wonderful cross-section of orgs!
As information theory teaches us: we measure success of communications by the RECEIVER.
top down = broadcast (one way) rather than communication (two way)
โThe single biggest problem in communication is the illusion that it has taken placeโ โ George Bernard Shaw
What prevents shared understanding? - too many buzzwords - lack of playback to ensure shared understanding - fear of looking stupid (I am stupid, I gave up trying to hide it) - what else?
@luca that is often the case when things are imposed on you rather than you contribute to the approach / solution etc.. (IMO)
If there is lack of interest, perhaps they should use the law of two feet and leave the meeting. I donโt have a problem with that, in fact it would reduce time wasted in meetings
But also: I've observed too many transformations where top management somehow concluded that while they endorsed it, it didn't concern tehm
@stevesargon maybe lack of interest isn't quite the right way of putting it. They didn't realise it was in their own interest to be interested. @nigel.buddโs suggestion of lack of shared purpose feels like a good way of putting it
as @sam793 said during the talk, often the problem is that the understanding is both too big for any one person or team, not just organisationally, but in a time/history perspective. we're so busy solving the problem in front of us that it's too much effort to go outward. and even if you did try, our remit os quite narrow, and you start to hit lots of roadblocks without really high-level authority
here's something i wrote on how much effort (and some things you can do) to get untangle it: https://blog.cambiont.com/post/the-practice-of-digital-transformation-four-non-tech-skills-for-introducing-change-and-innovation/
2I really liked the Pfizer talk that had that high level vision of curing a billion people, that's the kind of vision that you can get behind and see how your team can help reach that goal
Loved that pic from one the Watch Parties! (Oh, whereโd it go?)
https://devopsenterprise.slack.com/archives/C06UH7YSZU5/p1713956644864749
OMG. Itโs so easy to lose track of space and time โ I thought we had one more talk before break! Thanks @jeff.gallimore for keeping me on track! ๐
@jeff.gallimore @annp and @mvk842 have a tough job
1Manchester watch party! Ready for lunch ๐ great talks everyone ๐
Often die (Senior) Managers reuse the strategy slides they used for the shareholders / Board / whatever when talking to their teams. And - surprise! surpirse! - this is not necessarily a language the teams do understand!
@jaz.blakeston-petch original millennium falcon ucs set from 2005. my wife got it for me a few years ago. since it was used, it had the instructions, but not the numbered bags. so sorting through 5000 piecesโฆ ๐คฏ and itโs the millennium falcon. and the largest set at the time it was released.
Here's the Strategy to Execution Map that @sam793 referenced! Feel free to reach out to me, @chriscombe or Sam and we'll gladly help you work through it or better yet, run it as a team workshop ๐ https://miro.com/miroverse/strategy-to-execution-map/
for extra credit (once you have your first version), you can try to overlay your tooling landscape for some interesting perspectives / layers
โน๏ธ if you tag a post with a link or useful resource with the ๐ emoji, it sends that post to #C04ED43AQAC
Thanks for all the responses to the survey so far, you can see the results here: https://www.mentimeter.com/app/presentation/almat2m8synf5u8eufcpdhk7dnau6tq2 Itโs not too late if you missed out - https://www.menti.com/alxp9w1yrmf1 Having fun in https://app.gather.town/app/e2Hj0ENa9FLWluEr/ETLS sharing war stories
Reminder: The talks are starting again in 5 minutes. Start navigating your way back to your browser. https://devopsenterprise.slack.com/files/UATE4LJ94/F06V49GSELX/timer.png
โกPlease welcome Stefan Ostwald, Co-Founder & CTO, Parloa and @peter.petrovics, Strategic Advisor, Equal Experts, presenting How Parloa Revolutionized Customer Support with Europeโs Largest GenAI Conversational FAQ. They'll be reporting on launching one of Europeโs largest GenAI-driven conversational FAQ service for a multi-national enterprise, with 50MM customers in 30 countries. They'll talk about major hurdles, including those of โhalluciationsโ and privacy concerns, and how they overcame them.
The risk of corporate chatbots for customers: https://twitter.com/RealGeneKim/status/1763016504944292294
Also this... https://www.independent.co.uk/tech/chatbot-swears-ai-dpd-poem-b2481967.html
On the problem of LLM confabulations โ Iโm so delighted that Dr. Ethan Mollick is talking on some of this in the next session!
โฆand Dr. @mik will talk tomorrow about how prompts actually couple us to LLM, which can massively increase switching costs, during a period when we want to maximize optionality! (Which @internettitan will talk about tomorrow as well!)
There is a lack of standard metrics/benchmarks to measure how the behavior of AI/ML. Some attempts https://github.com/mlcommons/modelgauge https://github.com/mlcommons/modelbench
indeed. without good controls and guardrails (including humans), you get stuff like thisโฆ Facebook: two chatbots developed their own language https://www.independent.co.uk/life-style/facebook-artificial-intelligence-ai-chatbot-new-language-research-openai-google-a7869706.html
Lots of new sociotechnical wiring required to ensure good outcomes โ weโre beyond CI/CD and production telemetry.
wayyyyy beyondโฆ so many new elements and paths with this stuff
generative AI generating background noise to seem less fake.. so the customer doesn't think they are talking to a bot.. (when they are) wow some uncanny valley stuff going on
Agree, Iโd rather be transparent and tell the customer theyโre talking to a bot personally
Agree ๐ฏ, we are stating in welcome message it's an AI bot. There is still a lot learn what works well for Call Experience (CX) with more and more advanced bots (and more human sounding voices). Journey for providers and callers too. Though I think we are not too far from the point where it will be difficult to spot if I'm talking to a human or AI Anyway, the purpose of background sounds for us is not to "pretend" it's a human but indicate we are there -total silence actually sounds very weird. you feel the call hang up or the bot stuck.
Just looking... bringing up the documentation... sorry,the systems working a bit slow today...
actually I donโt want it to pretend to be human, that seems more uncanny
@peter.petrovics How much did you have to reorg dev, MLops, data science, etc, to do all this? What was biggest surprise?
We ramped up a team focused on this, almost greenfield (apart from call infra and a few other basic platform building blocks). The team started small but continuously growing with embedded data science etc. gurus. As we grow we are gradually "splitting" out platform / domain teams . Tbh I can't see delivering this much in such a short time doing it with an existing rigid team structure. As for biggest surprise: Good surprise: LLM capabilities still, up to this day can be jaw dropping, how well it can handle complex scenarios Bad surprise: maybe the level of LLM flakiness. Makes tuning, testing and evals very challenging
๐ซ And now, a special welcome to Dr. Ethan Mollick, Associate Professor at the Wharton School of the University of Pennsylvania and author of https://www.amazon.com/Co-Intelligence-Living-Working-Ethan-Mollick/dp/059371671X.
oh interesting problem - seems like a lack of safety if people are doing that behaviourally
Will Ethan Mollick address the comments about RAG from Geneโs LinkedIn Post https://www.linkedin.com/posts/realgenekim_i-had-a-wonderful-time-watching-james-cham-activity-7180796274537820160-_Dcr?utm_source=share&utm_medium=member_desktop
๐ก trendsโฆ โข specialized AI devices โข large context windows โข agents
"Wrote the policy to ban ChatGPT but used ChatGPT to do it"
We wonโt lose our jobs to AI, we will lose our jobs to people who know how to use AI (and how NOT to use it)
'Luddites' have been worried about losing jobs to technology for a long time! Early 1800s in England, smashing up looms. "Mill and factory owners took to shooting protestors"! Glad that doesn't happen now.
As a result we have no mills and little manufacturing left in the country, not entirely sure thats a win.
Have you looked at the quality of the documents we have? Will loading lots of low quality docs help or hinder?
this to get started: https://www.linkedin.com/posts/realgenekim_i-had-a-wonderful-time-watching-james-cham-activity-7180796274537820160-_Dcr?utm_source=share&utm_medium=member_desktop the comments might have a link to the actual talk.
Or IT blocks people from using it, until they have made it โsafe to useโ and they loaded all the proper docs and tuned it.
And we need to put the models in the hands of Domain experts not just IT!!! (Sorry ranting here)
We call that the "innovation tax" - it's expensive now and will change a lot, but we understand it better because we "worked" it
Thank you for all your sharing Ethan - it's how we get better as an industry !
Help decide where things are going by publicly sharing early successes
I heard from a procurement manager that said all his suppliers were saying how quickly ChatGPT helped them make their tenders and the procurement manager, yes but you're causing me to have to make sure it's rewritten correctly for presentation. What's a good way to handle this sort of situation?
โจ Next up, @jpetoff, Director, Google Cloud Platform and Technical Infrastructure Education. She will talk about what the Google SRE organization is doing to create a new talent pipeline, focusing on universities and new engineers.
PS: I loved Dr. Ethan Mollickโs book โ I think itโs such a valuable book for non-technologists and technologists alike.
on people being capable SREs without decades of experience? Impossible, right? ๐ @jpetoff
Its not just Education failing to properly educate SREโs The same also applies to CS in general.
that StackOverflow survey I mentioned: Source: 2019 StackOverflow Developer Survey: https://insights.stackoverflow.com/survey/2019#key-results
@jpetoff, when do you think Corporate outreach should start? Is it just from a grad level onwards?
@matt.pickles I think primary and secondary school outreach is also very important to keep people interested in Tech in general, but 3rd level (uni) is the right time to think about specialization in an area like SRE. Also, our program was driven by volunteers so it was important to scope it carefully (focus on uni outreach) vs spreading ourselves too thin
@jpetoff, I admit that's probably a better answer than I was expecting, so within the cohort of people who have already decided on a tech career you encourage them to look at an SRE career, that's quite understandable.
@jpetoff So what about applying the same approach to career changers, people who took career breaks for caregiving, raising kids, etc? There's already a pipeline for students, and there's nothing wrong with refining that pipeline. But perhaps expand that pipeline to include people who are not new grads but are willing to learn something new.
i bet @forrestbrazeal has something to say about thisโฆ https://cloudresumechallenge.dev/
I agree that this is also important. We are doing some work in this space (e.g., we piloted an SRE fellowship with Major League Hacking) and are currently running a course on SRE with Uplimit that tends to attract more career changers.
@jeff.gallimore, thanks for sharing the resource from @forrestbrazeal. Will definitely check them out. In the same spirit, I wrote this book here (https://a.co/d/3np4sw7) 6 years ago to share some tips as well for navigating careers challenges.
1@jpetoff, good to hear that you are working in this space. This needs more visibility than the opportunities for students. Ageism and other isms are real in IT and elsewhere. I find that there are plenty of people who don't necessarily have a tech background, but their disposition to learn, and their parallel experience in business makes them good candidates for tech work.
I gave a talk at GDG Galway recently which was sponsored by Genesys. I got to do a career chat with their interns while I was in town. I was really impressed with Genesys' commitment to bringing in interns from non traditional backgrounds. In this case, they had a strong pipeline via a program an NUI Galway. I met a former hairdresser, clerk at the UN, and more. So inspiring to see how they were thriving.
@jpetoff OMG. I WANT TO DO AN INTERNSHIP!!!!
@kareen Good point. Also career changers, who change within the same company already have a head start based on Domain knowledge. Example: A Bank, had downsized some folks and created a Project Athena type approach to train them as developers. The advantage being they already knew the business and the customers.
@stevesargon, yes, this is a good approach to repurposing the skills and interests of people. It's good for the company to improve operations, as well as bring along the people who helped them get to that level in the first place.
โall roads lead to SRE eventuallyโ :rolling_on_the_floor_laughing:
I love the focus on building positive ripple effects in any company. Even if not staying at Google the positivity still spreads!
Join Google and take our internal SRE EDU Orientation program ๐. How cool is it that I lead a program with a rainbow farting unicorn (RFU) as a logo?!
Gotta have a TLA (three letter acronym) to give it that extra bit of corporate cred :rolling_on_the_floor_laughing:
Wowโฆ this is so amazing how @jpetoff has solved for growing SRE skills
:satellite_antenna: Please welcome @akashrajendra_venteka and @qiheme, both first-time engineering managers from Comcast. They'll talk about why they decided to become a people manager, the challenges and learnings making this transition, and the resources they drew upon.
Thanks a lot for your support throughout! @annp @alexb @genek! Couldn't do it without you!!!
@genek I love how you have organised your books by color. ๐ ๐ My sister does the same ๐
Make no mistake, the colored sorting of books is because of my boss, @mvk842 โ when she said that she wanted to do this, I almost fainted. It just seemed so improper and just isnโt done!!! ๐
1@genek I guess your wife and mine share the same addiction ๐ Mine didnโt stop at the bookshelf in the living room, but she moved forward to organise by color all my Italian suites, ties and shirts. :man-facepalming:
Nothing wrong with a bit of beauty and order in the worldโฆ ๐
If you're at the London watch party, there's a book stand outside the auditorium! ๐
Said no one ever: โI missing writing YAML config files.โ ๐ But flow state from building things is amazing and real. cc @qiheme
So key to have a pathway back to IC! That's great that Comcast doesn't force you into a one and only progression path. Have to find the right path for delivering joy!
โI refrain from opining on PRs, because I think Iโll slow down the process. So I reserve that to the guild meetings.โ @akashrajendra_venteka
"it's just a question" not a command.... So simple yet so easy to not make clear.
Fascinating advice from @akashrajendra_venteka โ without a doubt, the worst health I had in my entire life by was during my first couple of years of people-management.
@qiheme what would you say are the best ways and best balance between learning new leadership skills and keeping up to date technically?
This is a tough one because this was a conclusion I came to with my own personal experience. For me, it was being vigilant in understanding as much about the technology as I could, without actually writing it. I would personally be a part of most all architecture meetings for features, along with peeking in at the code whenever I could. There's definitely a point where you realize you don't know everything (In which case I leaned in on my engineers) and I focused on what I needed to know to guide my team and relay important information to stakeholders.
Thank you @qiheme for the honest answer, in reality that's pretty much as I imagined it would be but good to hear it from someone with the life experience.
This is definitely the way - stay technical without being hands-on if there is no time for being hands-on.
Did you ever lead "mob programming" sessions so that your teams could learn by your best practices and experiences rather than with a purely advisory/guidance approach? @akashrajendra_venteka
Yes, its a working session! As a technical lead, I used to facilitate 30 minute guild meetings (we used to call them 'Council of Elrond') where we have gone through software books, best practices, engineering blogs, etc. together for the first 20 minutes and reserve 10 minutes at the end for discussion on what were attendees 'Aha moment' or what did they learn from it? Currently, we work on onboarding, ReadMe standardization for github, best practices, etc. We make a lot of decisions on the future of our tech stack and execution of items.
โโฆbecause, last I heard, this is still a technical fieldโฆโ said @schmark
โค๏ธ that feedback is criticalโฆalso how feedback is given and received.
"Remaining technical" Definitely:+1::skin-tone-5: Every day needs to be a school day, no matter what level of leader you become!
Corollary to โLeaders are not born, they are madeโ โGreat developers are not born, they are madeโ Favor Nurture over Nature
@akashrajendra_venteka, great ideas about self-care. Sacrificing your health is not required to show you're devoted. @qiheme, great to hear you made a simple list for what you wanted.
We are indeed having some awesome discussions over here in Manchester.
Hope you can share some notes/learnings from the watch parties!
As you get more senior, youโre expected to give back to the community
As uni $ goes up, value prop goes downโฆ so how to reach out to non-traditional potential folksโฆ
Start out with good literature in [cs] education (@stevesargon)
Reminder: The final talks of the day are starting in 5 minutes. Start navigating your way back to your browser. https://devopsenterprise.slack.com/files/UATE4LJ94/F06V49GSELX/timer.png
DevOpsDays Zurich talk by former tech recruiter is good, per @jpetoff
I was just thinking, I'm not sure if they recorded the Ignites. I hope that they did because the talk was great.
Welcome back from our break! We are now coming up on our last set of talks on Day 1. I hope you are liking this new live format for the Enterprise Technology Leadership Summit! If you like it, can you me a favor? Can you please post about what itโs been like on LinkedIn, Twitter, or whatever, to let people know what you think, especially if you are enjoying it and finding the conference valuable. Letโs use hashtag #ETLS. You have my genuine and heartfelt thanks!!! https://www.linkedin.com/posts/realgenekim_etls-activity-7188908241756585985-Qoys?utm_source=share&utm_medium=member_desktop
1Technical problems are easy, people problems are hard (@stevesargon)
๐ Next, please welcome Michael Vormittag, Head of SAP Delivery, Architecture, Analytics & CTO Office, Daimler Truck and @marko.klemetti, CTO at Eficode. They will be presenting: How a Major Organizational Shift Initiated a DevOps Transformation at Daimler Truck. They will describe a unified developer platform called The Truck Toolchain (T3), a platform designed to enhance Developer Experience and pave the way for future technical innovations, including GenAI-based tooling.
Is the spinout like a well funded and experienced startup?
shutting things off and consolidatingโฆ this is not trivial.
@akashrajendra_venteka and @qiheme From your talk - did you folks have a suggestion or recommendation on the "manager assessment tests"?
I haven't done manager assessment tests. But, I have taken few assessments such as Insights, Birkman and DiSC which was helpful for me to evaluate how I work, how I work with others and where should I make adjustments.
I have taken Insights and Birkman, both have been eye-opening in understanding more about myself as an individual
Ah, interesting. Thank you for the references, I had not ran into Insights or Birkman before! Over the years, I've been in groups that have taken the DiSC assessment, and MTBI, knowing that while each of these approaches have some well documented flaws, the fact that the whole group (15-30 people) took it at the same time, and we had a group discussion afterwards was incredibly valuable. The best tool I found for self-knowledge was from Gallup - called StrengthsFinder. What is a bit missing from that tool would be any indications of "how would person X be as a manager". Certainly anything that could give a person at least a self-evaluation of some amount of current manager capability and future manager capability would be quite valuable. Lastly, I wanted to reference Gretchen Rubin's 4 Tendencies model (Upholders, Questioners, Obligers, Rebels) is another lens to look at how you interact with people.
is it the big 5 (or another one) that is much more science backed.. I've seen MBTI proven as accurate as horoscopes ๐
Yes, I am looking forward to more assessments. Thanks for the recommendations! Adding MTBI, StrengthsFinder, big5, Gretchen Rubin's 4 Tendencies model to my list!
How do people tackle the challenge of balancing Team Autonomy with tooling standardisation?
Shifting from compliance & cost-driven standardisation to value driven standardisation.
platform engineering (treating platforms as products) --> instead of forcing these platforms, you make them so good and so easy that the alternative is harder / requiring more painful .. also autonomy in enterprise doesn't necessarily mean you can do anything you want - there are always some guide and guard rails.. that is still plenty of room to decide how you build / run
Indeed. If you have to force people to use the platform that was set up to make their lives easier, something has gone wrong.
@marko.klemetti and Michael: reduction of application landscape by 40% โ amazing! Iโm so curious which applications were the most challenging to shut down โ and what were you most proud of as a result of doing so. So good!
PS: so interesting how many presentations used AI โtext to imageโ!!! Iโve only used it handful of times, and only for personal stuff!
that is something I will be using more in the future too - some great images for sure
"AI to development is like Aglie to Waterfall" So that means everyone uses the word "AI" but nobody knows what it means?
or โ everyone adapts the definition to what they would like AI to be?
1every small straightforward algorithm is referred to as an AI today. in some sense, it might be true ๐
Love that Daimler team (@marko.klemetti and Michael) are referencing the Adobe presentation on AI governance coming up from @bscott and @dneff! Shows the commonality of our journeys!
โbig companies tend to build a strong [and often constraining/stifling] safety netโ
Need to move from coercive bureaucracy to enabling, as per https://itrevolution.com/product/the-delicate-art-of-bureaucracy
Just out of curious: by show of emojis, how many people are are being given responsibility for rolling out or governing new AI tools?
One thing that's really helping in our case is having a really good lawyer as part of the team
One of the best use cases of AI for development I've seen is the linting and correction of standards of a pull request before peer review.
One of the best use cases of AI for development I've seen is the linting and correction of standards of a pull request before peer review.
@matt.pickles what are your favourite tools for this?
I'm still working on it for my own interest but it would be some kind of copilot and custom prompts/scripts with GitHub actions along with the main Linters for example, EsLint for JavaScript to correct the code.
Sorry for the late reply @vladyslav.ukis but it's not really much more than I said above. it would look something like a copilot/AI assistant(LLM model with the guardrails of the enterprises private environment) coupled with the already existing automation tools such as Github actions (or similar) and Linting tools, the idea would be to keep both the developers original code and the corrected code to provide a learning experience either "physical" or automated within the development tools as a final "coup de grace" for the process. Some links that touch on this are https://docs.github.com/en/actions/automating-builds-and-tests and https://www.freecodecamp.org/news/what-are-github-actions-and-how-can-you-automate-tests-and-slack-notifications/
@vladyslav.ukis, in the time I've been looking to this a GitHub action 'luiyen/llm-code-review' has been released in the GitHub Marketplace which uses the "meta-llama/Llama-2-7b-chat-hf" through HuggingFace. I've also seen other use cases that use llama-2 for similar tasks so it looks like a good base model to start with for prompt engineering and fine-tuning. https://github.com/marketplace/actions/llm-code-review
how is software development really going to change? wowโฆ thatโs a big - and fun - question to discuss.
:adobe: And now, please welcome the team from Adobe: @bscott, Principal Architect and @dneff, Principal Cloud Architect. They will share their "experiences onboarding hundreds of AI use cases and a process that allowed us to move as fast as a developer with the certainty of a lawyer."
balancing responsibility, speed, interest, and risk :juggling: ๐คฏ
What an amazing set of AI announcements from Firefly!!!!!
โBased on onboarding experience with finance, legal, security we were appointed ownership of company-wide Al intake in addition to existing responsibilities.โ
span of concerns for governance: everyone from finance generating summary content, integrating third party content into our marquee products, tools used by devs. So cool.
โwe want onboarding with AI to be dull and routineโ @dneff
Balance โmaximum libertyโ and โmaximum responsibilityโ !!!!!
i love this framing with the tension โ liberty and responsibility
:robot_face: :robot_face: :robot_face: Regarding the Daimler and Eficode talk: Has anyone here built Retrieval Augmented Systems (RAG) for your Development Value Streams to use? And what kind of problems have you solved when you have built these systems that @marko.klemetti and Michael were talking about?
For me it's the going to be focus area for "years to come" ๐
@dneff whoa. Thatโs me. I use IntelliJ with GitHub Co-Pilot. (And the IntelliJ AI so it can write my commit messages โ which @jason.cox told me about, which is a life-changer!!!)
(regarding AI written git commit messages โ I was on a plane without wifi, and AI couldnโt connect to mothership. It was terrible โ I had to MANUALLY WRITE MY OWN COMMIT MESSAGES!!! It was awful! Like I was in the Dark Ages again!) ๐
i love the clarity of this framework and the consistency it provides.
@bscott โyou may have other stakeholders you may need to loop inโ. Good caveat. ๐
For those that like to use cli is there a good solution to use something like copilot for commit messages?
Have you tried the copilot cli? https://docs.github.com/en/copilot/github-copilot-in-the-cli/about-github-copilot-in-the-cli
This looks like copilot for cli, I wanted something more integrated similar to what happens when writing ode in the ide but that would write the commit message.
Yeah, you are right, I don't think it has the commit message as of yet (but probably does in the future). There are a few initiatives to this: https://docs.gitlab.com/ee/user/project/merge_requests/ai_in_merge_requests.html (not cli, but kind of what you are looking for) https://dev.to/bdougieyo/ai-generated-git-commit-messages-4j7g
well there's a generic tool to interact with LLMs wich is called, unsurprisingly, llm which should be suitable for this purpose.
If you're thinking specifically of commit messages, I suppose you'd have to integrate it with git manually, but I guess that would just be a very slim wrapper or even alias
Clearly they did not have the power of today's Photoshop to create that poster... ๐
I love all these references to Wiring the Winning Organization: ๐
Seriously, I am blown away by the fact that once again, this community is on the vanguard of some of the most important technology and business initiatives! ๐
We focus on shortest job first, which usually means lowest risk proposals go first.
Fascinating: AMPLIFY: Broadcast out the highest value use cases, and the the low risk uses cases, to accelerate adoption of great stuff, and help people get their proposals thru. Marvelous!!!!
"A little bit of human sugar interaction doesn't hurt." One of the best lines of the day!
โญ Now presenting The Top 3 Patterns From Past Ways of Working You Need to Know! is @jonathansmart1, Founder, Business Agility Coach and Leader, Sooner Safer Happier.
We don't study the past in general, especially in Computer Science. ๐ข
those who do not learn from history are doomed to repeat it
and those who do study it are doomed to watch others repeat it
Q: do you think software should be in the same bucket / group as AI> or different paradigms (of industrial revolution)
@mik in his diagram made a new row for AI โ personally, Iโd add AI to the Age of Digital, which is how Dr. Perez does it โ and it underscores how long it took for us to get to the new normal โ over 2 decades!
(PS: what a great idea from Dropbox Capture โ name your file after your tool, so everyone knows when screenshots come from it. ^^^^)
I find it interesting that essentially all characters introduced were either British or American. What's the reason?
Centers of the Industrial Revolution? Just like most characters of renaissance were Italian I would say
What about primitive societies where there was no "leader" only "situational; leaders?" Leaders for hunting were experts in hunting Leaders for food were experts in gathering and gardening ...
I'm not sure their cave paintings were clear enough to explain their thinking ๐
Fascinating โ on Industrial Revolution. I think the latest Nobel Prize winner was awarded because of her work studying how embargoes in the 1800s protected French textile industries. A fantastic natural experiment showing effectiveness of industrial policy, which most economists dismissed as mostly ineffective.
Or as we Americans popularly pronounce, โcar-NEGG-eeโ. (learned from watching lots of public TV documentaries. โThis show was made possible with the generous support of the Carnegie Foundationโฆโ. ๐
Gene knows this from being coerced into watching Downton Abbey.
Tomato, Tomarto ๐ So many ways to say Carnegie! CarNEGee, Car-NEE-gee, Car-NUH-gee, Car-NAY-gee
Car-NAY-gie apparently https://www.wesa.fm/arts-sports-culture/2019-04-18/car-nuh-gie-on-car-nay-gie-depends-on-where-youre-from
Upon reflection Iโm now more conscious of and have empathy for the Brits when I go to England and canโt speak โEnglishโ
Itโs funny โ where I grew up, we said CAR-negg-ee. Which no other parts of the country said. The Public Broadcasting Service (like a snooty BBC) that aired Downton Abbey and created shows like Sesame Street would always say car-NEGG-ee.
To paraphrase Richard Feynman/Shakespeare/Eric Evans > whatโs in a pronunciation, as long as we have a shared understanding of what it means
Although, one does have to be careful. My brother spent a college semester in England, and a friend stopped by and asked โdo you want to go over and knock up Eleanor?โ Which has very different meanings in the US vs England.
Or if youโre from western Pennsylvania, you pronounce it kaar-NAY-gee
1(Rockefeller Foundation another huge funding source for American public TV!).
Random question, Is this chap related to โHow to win friends and influence peopleโ Dale Carnegie?
This is an amazing American history lesson!!!
Steve Jobs must have learnโt from him on the no-note taking Under cross-examination, Mr Schiller said Apple behaved much more like a start-up than a big company, with categorically no note-taking, almost no presentations, very little written analysis of the business and very little talk of profits in the executive team meetings it holds every Monday from 9am until noon.15 Apr 2024
we were just talking about boxes and lines earlier - some jobs haven't changed that much
I guess itโs the same for Greeks with Romans, and then for Romans with UK ๐
another fun learning is Fordlandia that Ford tried to do in Brazil (super interesting experiment - that didn't quite work)
This resulted in an 8x improvement in throughput. Replicated at the Olds plant.
McCallum also wrote (not very agile): Six general principles of administration (1855): โข A proper division of responsibilities โข Sufficient power conferred to enable the same to be fully carried out, that such responsibilities be real in their character โข Means of knowing if such responsibilities are faithfully executed โข Great promptness in the report of all derelictions of duty that the evils may be corrected โข Such information, to be obtained through a system of daily reports and checks that will not embarrass principal officers, nor lessen their influence with subordinates โข The adoption of a system, as a whole, which will not only enable the General Superintendent to detect errors immediately, but will also point out the delinquent.
Awesome talk @jonathansmart1! Good to see you again.
Yes, thank you for feedback, and thank you for posting on social media on how the day went!!!! https://www.linkedin.com/posts/realgenekim_etls-activity-7188908241756585985-Qoys?utm_source=share&utm_medium=member_desktop
I love who studies history to understand the present. Amazing work as usual @jonathansmart1 โค๏ธ fantastic regression rolling back the timeline to understand how we arrived to todayโs way of working
Thank you to all the speakers!!!! What an amazing set of achievements and learnings!!!!
5and holy cow - @genek "the machine" types notes faster than anyone I've ever seen - need to get him to do a CAPTCHA test to make sure he is human ๐
@jonathansmart1 Top Trumps should definitely do an industrial revolutions set based on your presentation ๐
Rebel: 10 Temper: 2 Patience: 11 :rolling_on_the_floor_laughing:
Love the single-threaded format, helps prevent FOMO and creates a shared experience scenius for us all as a cohort
Would suggest leaving 5 min open between talks for the speaker(s) to review and respond to Slack questions while still โon airโ. Hopefully thereโs more ability to be flexible with this formatโฆ
Looks super fun!! If FOMO is the Fear Of Missing out, then what acronym described the reality of missing out? :thinking_face:
AMO - actually missing out? Or, how about SAMO - sad about missing out? I think we could start a new trend here if we play our cards right :-)