This page is not created by, affiliated with, or supported by Slack Technologies, Inc.
@joshua.andrew.phillip posted this, describing how his conference/agenda app is now in App Store pipeline — in genuine awe that he got this app written so quickly. Kudos! https://www.linkedin.com/posts/joshaphillips_i-came-to-the-enterprise-tech-ai-summit-activity-7376774620797902848-hADq/ cc @steve.yegge
Thanks for the reshare Gene! All credit goes to you and Steve for the inspiration and your workshop serving as the catalyst! Hopefully I'll be able to drop an app store link today 🤞 Side note for everyone else: I called my wife this morning sounding like I had just met Luke Skywalker because Gene Kim had commented on my Linkedin 😅
@philipp.goellner Holy cow: https://www.linkedin.com/feed/update/urn:li:activity:7376534446910521344/
I noticed the trains are out at the airport. They are saying allow at least 3 hours. Might want to make sure everyone knows this
Get yourself to your seat in the Royal Ballroom for the opening remarks. We’re kicking off the final day of the Enterprise Technology Leadership Summit in 15 minutes at 8:45am PDT! https://devopsenterprise.slack.com/files/U06GCH026KT/F09GAC5FBEJ/timer.png
Remember all those talks you attended the first two days of the Summit? Please submit your feedback for those! It’s so valuable for us and the speakers. And after all, feedback is a gift and sharing is caring! Enter your feedback for those talks here: https://etlslasvegas2025.sched.com/ https://devopsenterprise.slack.com/files/U06GCH026KT/F09GFGYPA0L/feedback.png
We want to hear your stories from the Summit. What did you learn? Whom did you meet? What ideas are you taking back with you? What actions are you planning to take? Post in #CB0936XFT! https://devopsenterprise.slack.com/files/U06GCH026KT/F09G8DU6LBF/stories.png
It's amazing how approachable and engaging all of the speakers are here! It's so awesome
Awesome week @ #ETLS25 Great sessions. Looking forward to next year already.😊
let's keep the slack active throughout the year! I love this community
☀️ Please welcome @mgill, Sr Director of Engineering, Core DevOps, GitLab, here to present: Building (and Keeping) GenAI Teams
AI features, model validation, the infrastructure it runs on, scouring advance— what did she do right or wrong to get put in this role? 😂
Amazing week!!! First time here. Learned a lot and met some awesome people!! Looking forward to next year @#ETLS25 thank you program committee!!
This applies to you if your HR manager comes running to you: "Hey, I need an AI engineer (whatever that is)" 😂
First time at the conference and have been blown away by the quality of conversations where folks get real about the challenges their facing, what’s working, and what’s not… Love the willingness to be transparent and share. ❤️
Love that you’re seeing this. It’s a hallmark of this community and one of the many things that makes it so special.
Michelle just reminded me. I forgot to mention the most important thing about options as it relates today: when uncertainty is high (like in AI, where no one knows anything), do not make long term plans (which deprive you of options). <----
Commit - feels difficult with so much change happening right now, but a good reminder to go back to basics “Working Software over …. “
🧑💻:skin-tone-4: Next, please say hello to @jbeutler, Head of Solutions Engineering, Strategics, OpenAI, presenting: Making AI Agents Actually Work for You
The LOOP! For many, the loop is the most important characteristic of an agent. (We used this definition in the Vibe Coding book)
What's hilarious to me is that previous to OpenAI, @jbeutler was a solution architect at Stripe — for him to talk about "lack of credibility" is so interesting to me. Sure, he wasn't writing production code, but still! (It shows the unspoken hierarchy within the engineering community.)
Again, the observation that team sizes are shrinking — who can go further and faster.
🚗 And now, please welcome, @david315, SVP of Product & Engineering, Cox Automotive, and @internettitan, Field CTO, Boomi, here to talk about From Technical Debt to Digital Credit: A Case Study in Serial Optionality
Another #VibeWithYourTribe moment from a first time attendee. The flow in my conversations has been incredible this week. It feels like there is no handoff tax - just pure throughput of ideas.
--> When there's an earnings miss, what happens? "It's common to cut everything that gives you choices, that gives you options in the future."
How do you balance the cost of building with optionality vs the speed building with specificity? Any clear tradeoff frameworks folks are using? I know we can do things like assess risk of modular pieces being replaced like typical, but what are some of the ways you guys look at that differently or how do you decide how much optionality makes sense? Looking for less art more science type ways of assessing and implementing
AI is making it easier/faster to run parallel experiments to try out and narrow down options to the most promising.
Full disclosure I had not seen the presentation content ahead of time. 🙂
We make it easy in the ooops framework for teams to explore and align on this
• Formula for optionality: (N × K ÷ T × σ) ◦ N = number of modules ◦ K = parallel experiments ◦ T = time to experiment ◦ σ = risk/uncertainty … a key point here is that you don’t really need to fully quantify uncertainty… you only need to know directionality and order of magnitude because you’re only looking at it as a decision making tool to guide your investments in the composable set of capabilities.
Is anyone else impressed that Gene was able to type this on the ipad touchscreen live on stage? Including finding the σ symbol???
So is the goal of this formula to maximize optionality given constraints of K as teams or people that can experiment, and T being constraints that we can afford to let them iterate?
Interesting that lowering risk increases optionality given constraints in this formula.
Set the defaults to favor optionality and Maximize them in contexts where uncertainty is high (while working to lower the cost of experimentation)
A quote I seem to remember from somewhere: you can debate the accuracy of a number, but when you're working with orders of magnitude then the specificity isn't all that important anymore
We talk about this all the time in the context of flow engineering. We don’t need to know whether cycle time is 2 days or 3 days for a stage when there’s a stage that takes 8 weeks https://www.linkedin.com/posts/devopsto_a-director-of-quality-assurance-asked-if-activity-7376653962290270208-p5w9?utm_source=share&utm_medium=member_ios&rcm=ACoAAAIORJIBEoHqn58LQs7XJfr3BzzXs9n47cU
Okay second question - for smaller teams that can’t afford more parallel modules, or where more modules has complexity of maintenance, do you have any formulas for balancing how much modularity you want before you start to have issues with maintenance burden and context burden?
Agentic SRE functions will reduce the size of this problem. For now I think you find your Σ
Interesting thought, so if we build more optionality it probably would make sense to build agentic context as a feature of the module so that you CAN enable agentic SRE…
@bubthegreat - this is interesting to unpack… I’ll choose to skip over the cost/benefit part given that it’s a long conversation and instead get to what I think is the intent of your question… “how do you pick your spots?” That’s the second part of the OOOps methodology spelled out in UTE - optionality, OPPORTUNITY, optimization We lay out an easy toolset and pattern book for teams to learn how to map it out and align on where to invest (and how to set the defaults)
Like @jbeutler said…products will start to adapt to the user vs. the other way round (as it becomes easier and easier to add variance)
The implication I’m hearing behind that is that for it to be successfully “cutting the dead weight” we are using that user feedback to actively kill dead branches of modularity that haven’t played out successfully? Just making sure I’m inferring the right expectations - so if my SDLC doesn’t actually include iteratively killing modules that haven’t played failed, that’s on me for having unsustainable tech debt balance issues?
Yeah I think so…and that “dead weight” can be identified through objective data collected along with user (subjective) feedback.
@josh.guice - this is exactly right.. it’s the pace of adaptation that’s changing as the cost of software continues to drop. And… There’s a methodology and pattern book spelled out in the book to make it applicable and practical. - The OOOps methodology - Optionality, Opportunity, Optimization
This conference is a life changing event for me. I am so thrilled to include vibe coding in my daily life. Thanks to this community which is so enthusiastic, knowledgeable and caring. I feel rejuvenating my career with confidence and support.
I felt the same way about this Conference last year at my first attendance. Great conference and great community!
Meaning outcomes and clear definitions of success are even more important.
1Optionality should now be part of your core table stakes for software design feels like “design with agile iteration as the expectation” - which I love
Would anyone like to buy some of my technical credit? It’s for sale, and I will give it to you at a hefty discount to make sure it goes to a good home
My leader said it’s time to rehome it because it’s not working for us, not because it’s inherently a bad technical credit :rolling_on_the_floor_laughing:
Feel free to connect with me to discuss it more along with how to practically create/conserve it
I’m interested in the potential of making it a currency that you incentivise - so that it’s truly balanced against conventional metrics
💡 Please welcome @mik, Author of "Project to Product" and CTO at Planview. He will be presenting: Output to Outcome: An Operating Model for the Age of AI
I heard yesterday from someone that one team is now building so many features, for the first time ever, teams couldn't finish all their feature demos during the 1 hour meeting — big questions ahead!!! More demo meetings? Don't demo every feature?
Source for those unfamiliar with the reference: https://medium.com/sooner-safer-happier/agility-build-the-right-thing-69d316aeb56b
It’s honestly incredible to me how much success banks on changing the mentality around how to approach things and how much harder that is than just implementing the ceremonies and artifacts, and how often agile transitions don’t focus on changing that mentality
Yes, and... There are benefits to acting to your new way of thinking. https://medium.com/10x-curiosity/act-your-way-to-a-new-way-of-thinking-john-shook-d7df580fb10c The anti-pattern is implementing ceremonies in ways that don't actually change how we behave or influence how we think differently.
@stuart.pearce mentioned on Day 1 on how so little of the SDLC is code generation — suggesting something is very, very wrong.
This is a great highlight for why so many isolated agile adoptions fail. You can’t be an agile team if every part of the value stream around you is waterfall.
Outcome and System focus are the perfect antidote to hype 🙌
As @gnuyoga said, "for the first time in my career, the business and operations and support and HR asking me to slow down. Versus being blamed for being late, underdelivering, and going too slow."
Our organizations constraint is being able to state the business problem or requirement that needs to be met for the shiny new software they want to experiment with. This resonates greatly to our current state!
Don’t confuse activity with output or output with outcomes
I think a piece downstream of outcomes in Mik's model is "impact" which is generally the business impact that results from achieving (or missing) those customer or employee outcomes. The lagging indicators. Often quantified in revenue, cost management, risk mitigation, etc. https://blog.crisp.se/2019/10/16/christopheachouiantz/output-vs-outcome-vs-impact
A head scratcher: have you seen how many functional specialities are required to ship these new capabilities? It's like the hospital emergency department: In WWO book, we wrote that EDs were so much safer in the 1950s, because there were only 2 functional specialities (doctors and nurses). Now there are hundreds of functional specialties, far outstripping the Layer 3 ability to coordinate. Establishing flow in these systems require a super sophisticated Layer 3 management system. TLDR: leader's jobs will get more harder, not easier.
Complexity in healthcare is 100% driven by regulatory change. Almost every physician I talk to would swear that paper charts are faster. Classic automation for the sake of automation.
So often the failure occurs before we even begin by not having a strong thesis for why we are building “we believe building x will lead to outcomes y such that impact z”
Yes, hypothesis with experimentation and data feedback loops of leading and lagging value measures. Amplify what works and dampen what doesn’t. Complex Adaptive Systems are unknowable.
"We don't have time to define a thesis.. this takes too much time, boss said we need to replace x with y next week."
The "we believe" or "we bet" language also requires culture shift from deterministic to emergent thinking. Requires considerable psych safety in place (modeled by leadership) to be willing to be wrong (i.e. intelligent failure) and not be punished for it.
Hence the importance of ‘elephant carpaccio’, minimizing time to learning. And leading & lagging measures.
OTOH, we've also heard from many that AI reduces need for functional specialists (e.g., need something done in Drupal, Python, Neo4j, MySQL? Just ask the AI. No need to get a specialist.)
This org change is so right, and constantly demonstrated by conways law, but so hard, especially when my biggest stakeholders are the CFO and CDIO!
Diluted, shared ownership -- also highly correlated to what I call the "rockem sockem robots" management model
Operating model and The value stream network seem to be the least popular and simultaneously most impactful aspects of organizational performance :thinking_face: The answer I hear from nearly every leader I speak with is that “nobody owns that”
Implication of Conway's Law: if you have a complex shared resources in a functional organization, you have lots of dependencies and low cohesion/coherence. The countermeasure: duplication, which seems wasteful, but enables independence of action and high coherence
Amazon does a great job of making that trade off explicit and intentional, with clear guidance to prefer duplication over dependency or premature standardization 💪
This was highlighted in a very real way in “Teams of Teams” - creating duplicate missions at the edge but all aligned to a higher common objective
I think Mik is making the case that in the future, we're going to have lots of modular structures, or lots of duplication -- because that is needed to liberate teams to create value faster. (Many software architects may find this distasteful)
https://devopsenterprise.slack.com/archives/C015DQFEGMT/p1758668134563919
I see this every day with my clients. The goal to centralize is not the way. Image if the internet was all centralized.
The duplication within the org lets you run more parallel experiments (of entire paths to a target objective). K = parallel experiments
📖 And now, a warm welcome to @cbhawthorne, VP of Product Operations and Head of Technical Product Management, Datavant and @cbaldwin, Professor, Harvard Business School. They will present: Design Rules in Practice: Building Shared Capabilities at Datavant
There are a surprisingly high number of integrals and square roots and exponents in the option value formula and how it's used. (Which is why, I suppose, Merton/Black/Scholes won a Nobel Prize for it.)
I was trying to google this and keep getting directed to stock evaluation for it - does anyone have a link for applications in software specific examples or is it just generic application of the formula for a similar “value” assessment that they used for stock options pricing?
Hey @bubthegreat - I hear you.. the formulas/calculus are dense… I tend to think of this in a different way. I use @cbaldwin ‘s formula as a decision making tool more than an accounting tool. I.E., to follow the path of small bets placed wide in a sea of high uncertainty, you can set the defaults in your approach to make the happy accidents happen predictably through sequential optionality … one way we talk about it in the book is to use the casino frame. Key point to remember - casinos don’t gamble. They work the math because they know how probability works… no bet is guaranteed but “small bets placed wide works in their favor because they have the edge over the bettor” it’s not mathematically possible for them to lose because of the scale With tech options, it’s bit different and the bets are longer term in nature. While you’re not guaranteed to win just by having the option … you’ve improved your propensity because your cost/time to capture market share in a time of change is shorter than competitors who’ve taken the “shortcut by default” path that’s so pervasive in enterprise IT where the normative behavior is to compress teams and cut approaches that don’t have a deterministic outcome. I.E., it’s hard to justify decomposition for use cases you don’t know about yet… but that’s the point. The ones who are most ready to capitalize when the future arrives will have the best chance to exploit it. You can see it play out this way with AI in the last few years… so many of the digital giants were able to embed AI throughout their platforms in wildly short times… they could do this because the platforms themselves were both decomposed and built in the context of externalized use -> they built up technical credit and cashed the chit in when the future arrived
Please check out unbundling the enterprise… it’s more than the academics of it. It’s got concrete patterns and methods to help you apply it … all based on real world case studies across a wide array of industries. It also has a leader/practitioner guide section to help you keep your teams on the rails with clear and concrete risks and mitigation strategies so you don’t have to learn from your own mistakes… you can learn from the hard won experience of others who’ve done it at scale
https://itrevolution.com/articles/option-value-in-development-manufacturing-dora-metrics/
For what it’s worth, I’m not worried about the formula density just making sure that I understand the conceptual application
this is a master class in clearly explaining the key topology issues I see everywhere at large enterprises, amazing explanation. This is giving me such clear language to use to describe these things back in my own day to day life
Related to Cynefin too: https://en.wikipedia.org/wiki/Cynefin_framework
From the person who literally wrote the book on this. 🙌
The book (design rules) is more than a writing/creative effort … the research, analysis, and synthesis are stunning in how well it explains and proves how optionality works… it’s not theory, it’s observable reality with so many enterprises & products
Such a fascinating set of abstractions from Dr Carliss Baldwin. I love how clearly she explains how many things can be the same 'kind of thing' over time.
This is what got John Rauser and I so excited when we wrote Wiring for Flow. This concept of the “spanning layer” as platform, a constraint that productively connects the infinite variety of producers/origins to consumers/destinations (e.g. Internet Protocol is that constraint, and allows for optionality. The spanning layer on the internet that allows for effective operation/flow across all that variety) https://itrevolution.com/product/wiring-for-flow/
The slide deck for my AI for Managers presenation. https://docs.google.com/presentation/d/1JHOX8rtAzLJXW_o6bqZIbjeBAFv63GlB/edit?usp=sharing&ouid=109548076066236396148&rtpof=true&sd=true
The breakout sessions are starting in 5 minutes. Start navigating your way to whichever session you’re attending. https://devopsenterprise.slack.com/files/U06GCH026KT/F09GAC5FBEJ/timer.png
John Willis: "I'm just gonna scare the heck out of you for the next 30 minutes." :rolling_on_the_floor_laughing: Buckle up!
If I remember right, @topo.pal , while at Circuit City, was on the team that created the JAR file format for Java. Must have been early 2000s?? Amazing! (This must have predated and enabled the creation of Maven?)
I sure hope @botchagalupe closes with some kind of warm fuzzy! :rolling_on_the_floor_laughing:
I too wish to live in a fairy tale. 😉
Welcome to this brave new world. There is no going back. (Had Claude Code corrupt some of my data -- luckily, a friend told me I need to make a "one command to backup production database" command. Restore took an hour.)
Would love to learn more about the solution you deployed here Gene! Something that I'm thinking about as we try to set up our devs to do more vibe-coding (with the right guard rails in place!)
(I asked it to fix the query, but it changed the data instead. 😂 😂 I now automated the database restore process. I'll never go back to the old way — beats coding by hand!)
great talk from @botchagalupe this morning called "AI for Managers" but was all about "AI and Security (or a lack thereof)" - John, any chance to have a list of all of the reports, studies and other resources you highlighted throughout? That list alone would be an amazingly useful resource.
The breakout sessions are starting again in 5 minutes. Start navigating your way to whichever session you’re attending. https://devopsenterprise.slack.com/files/U06GCH026KT/F09GAC5FBEJ/timer.png
Any other Obsidian nerds attending? I've had some fun using the codex CLI to traverse my vault for all my notes about the conference, traverse the wiki links I'm creating, and creating a summary of the major themes from the conference (so far) based on my notes. It's so fun! I remoted into my mac mini at home to run it so I can have the agent running while my laptop is asleep.
Tom, look up my SDM "Marc Rawji" on LinkedIn if you want to geek out on Obsidian. He has contributed to it and get kudos regularly.!
Hey @tommertron would love to chat about Obsidian and what you’re doing with it, how you structuring your vault, what you’re doing with AI, etc
Always love talking obsidian! Happy to meet in the hallway outside the main ballroom after this talk (~230).
How do we allow users to accept the changes in our software at their place, not ours?
“We bought all the tools. Because more tools are better, right?”
"With all the fun dev tools, we got a 16 day improvement in lead time. Awesome, right? But that was only 6% improvement."
"My dad has an awesome set of tools! He can fix it!" - Jeff Spicoli, Fast Times at Ridgemont High
Could we get the example case studies from the slides after this presentation? This is going to be so useful for our organisation and AI adoption journey ahead!
Love this quote about this conference “My professional Disney land exists!” Thank you for letting me share it @geraint.levan
The final plenary sessions are starting again in 5 minutes. Start making your way back to the Royal Ballroom. https://devopsenterprise.slack.com/files/U06GCH026KT/F09GAC5FBEJ/timer.png
> Welcome to the last segment of programming for the conference! Some of you may be wondering what we have in store for you, and I can promise you that there will be things that are novel and new. > But before then, I want to share what an incredible 2.5 days here it has been for me at this conference, and I so much appreciate all the fantastic conversations I’ve had with so many of you. > But what I find even more exciting is that you are connecting with so many of your fellow travelers, not just to communicate but also to collaborate. > Since the beginning of this conference in 2014, people have asked for ways to continue these types of conversations — we’ve done some experiments, but something seemed missing. > But several of you asked for a structured way to cohorts of you to meet regularly, say, monthly for 90 minutes, between now and the next annual conference. It would be a combination of open conversation, facilitated discussion, and guest lectures from experts. All under Chatham House rules, where people can share and use information from meetings freely, but without revealing who said what or even who attended. > This seems so potentially valuable now, when the field is changing so quickly, and there are so many experiments happening that we can collectively learn from. > If this interests you, I’d love to explore this with you — just please respond with an emoji to this Slack message.
🛠️ Please welcome @ericc, Director - Engineering Effectiveness, and @philip.sears, AI Enablement Lead from Grainger, here to present: When Vibe Coding Doesn’t Vibe: Hard Truths in Enterprise AI
Afraid? Just because it deleted some data from our production database? That's what backups are for! Onwards!
Can someone from Grainger share the "Sensible Defaults" -- I would LOVE to take a look!!
I will look into if we can share that publicly. They will largely look familiar to this group I think :-)
🙏 🤞 (we've been trying to codify ours, but it would be amazing to have a place to start from)
"As an engineering leader I am able to express my views with prototypes and PRs instead of powerpoint" 👈 This!
Don't take a screenshot like I did of that list of AI vendors, like I did — those company names were AI-generated, to comply with some corp communication rules about not naming real vendors. 😂
Both safe and fast. Finding ways to operate within the beautiful balance of optimism and realism.
"No one tool to rule them all" -- if you are in enterprise sdlc tooling then limiting your platform to just the one way will quickly alienate the customers you're trying to serve
Super interesting — they're underscoring that when there's lots of uncertainty, don't make long term commitments.
Taking a step back to look at the value stream before applying AI means implementing solutions where the most impactful problems are 🙌
Grainger - ya'll have all had a great story all conference. Thanks for all your time and energy coming here to share your journey throughout all the 3 days! Appreciate your investment in the community. 
ETLS - The Conference for the Ones who get it Done [Powered by Grainger]
Custom rules in a simple .md has been an awesome win for us too... almost too easy but works. ➕
Grainger sensible defaults (10) Build for production Build in security Trunk based development Test driven development Pair programming Team ownership of quality Automated deployments Continuous Delivery Fast automated builds Smaller stories
3Only cost centers brag about cost savings — the reward is budget cuts. Play stupid games, win stupid prizes. 😆
"Surveys aren't enough" -- Producing sound surveys is much harder than most would expect. Just ask the State of DevOps folks 😂
“Meet them where they are” is perpetually great advice 🙌
Gotta do it! It turns out AI can’t do any better than humans when they get 5 copies back from Confluence.
:shopping_bags: Next, please welcome @rshoup, SVP Engineering at Thrive Market, here to present: How We Doubled Engineering Productivity at eBay, but Still Didn't Save the Company
"Context is key" - Isn't it a little sad that we seem more willing to have good documentation because it helps us AI better... but our new engineers have needed this for YEARS!
The most quoted person in the DevOps handbook! Thanks for reminding us. I had forgotten that amazing bit about @rshoup
I loved Grainger’s talk — It was a great reminder for me that every org and every person is on a different learning journey. While it’s exciting to see so much buzz around vibe coding, it doesn’t mean everyone should (or even can) jump in right now. Some people are ready to experiment; others may need more time, safety, or context first. I love that this community encourages one another, shares what we’re learning, and remembers that growth happens at different paces for each of us.
1"I also owned the platform organization" -- the value of having a strong executive champion to bring resources to bear to address problems
Love the Q: if you needed to deploy every day, what would get in the way? >> impediment backlog creation
Bryan Finster shared this with me the first time I heard it: “What is preventing us from deploying to production today?”
"I had no original ideas during this period at all. We just followed the DevOps playbook." <-- While I do not believe the bit about "no original ideas", I do love executing via deliberate practice of devops principles. 🙂
Mobile app: 2010: By 2024, they had become big balls of mud.
Yes. Love the way he said it. "Very much NOT modularized. Very much balls of mud."
> And we automated what we call a patch pipeline, where we could consistently do infrastructure upgrades, security patches upgrades dependencies, etc. Interesting! 💪:skin-tone-2:
Awwww. That's not a fun place to be, where you don't believe things will change. "He didn't fight me. He just didn't think it was possible."
"Every user-facing change met with near-revolt ("Seller Straightjacket")"
“This is how we’ve always done it, why change?” — never been a fan of that risk-averse, innovation-killing mindset.
Randy summarizing such complicated dynamics in monolithic planning in 4 points is amazing :star-struck:
It seems like the unit of work for metrics should have been the coordinated release more than individual services.
anyone else wish they had as much energy as Randy after 3 days of conference talks!?!?
I wanted to applaud, but didn't want to appear to be rude... It just seemed wonderfully epic.
Winning at the micro scale =/= winning at the macro scale :melting_face: See: winning at output =/= winning at outcomes
What’s most interesting about this to me is that AI helps with none of this and will actually make it all worse 😬
Agentic human factors and psychological safety 😂
Don’t know whether I should show this talk to my clients, or if it will completely scare them. But the first step to recognizing you have a problem is knowing you have a problem.
This is the sort of real-talk that orgs should be having challenging conversations about
LOVE this talk from Randy! (except for the fact that he's describing some behaviors in my org.)
I consulted into eBay (on agile adoption) back in ~2009. I'm so surprised/ not-surprised that nothing changed in so long.
I have definitely worked for VP X and survived just to watch them get fired. Honestly even when you survive you're pretty spent on the organization at that point (even though my next supervisor was great)
🎧 And now, please say hello to @michael.s.winslow, here to present: Delivering Music and Spoken Word at One of the World’s Largest Tech Companies
Shout out to the lighting designer in the back of the room who must have been waiting all summit for that moment to show off their skills (could they use AI though??)
I haven’t expected all of us listening to Kendrick Lamar during this summit :rolling_on_the_floor_laughing:
@genek I am thinking your connection with @michael.s.winslow has been underutilized for after party entertainment at ETLS. :rolling_on_the_floor_laughing:
"Once we realized we were being measured" -- people will always game metrics. The trick is to drive maturity when people do it
I love this idea, great way to build a digestible narrative! I wonder if it can be transposed to other media "Make all the characters members of Monte Python's The Holy Grail"
@michael.s.winslow I would be interested in your waste identification algorithm... I hope connecting as humans is a value add activity? 🙂
Oh this is definitely not waste. It would likely fall under "necessary non-value-add" at worste.
Bringing us to a close will be @jonathansmart1, who will present: Certified Really AI Practitioner (CRAP)
we need to get this talk over at Enterprise Platform Tech at Capital One, it would kill
can we crowdsource John Smart to make an Office Space / IT Crowd style movie based on this talk?
SO much learning and meeting SO MANY amazing humans. Thanks, everyone!
Hard to believe we’re already at the last day of the summit! Grateful for the incredible sessions, insightful discussions, and most of all, the chance to connect with such inspiring leaders. Looking forward to staying in touch and carrying these learnings forward.
Not sure where to post this question, so just going to go for it… Our security experts are concerned about the security around MCPs. How are you securing your internal MCPs? Are you using particular vendors or technologies? When using external or 3rd party MCPs, do you have similar concerns around the security/validity of the information exchanged?
We’re evaluating vendors like Portkey and Cequence who are thinking about this and helping manage user level authz between agent tools and systems. For unblocking quickly - you can vibe code a simple server that integrates with your identity system and can provide some basic tools to agents while you are working on a more scalable enterprise solution.
@chuck - that’s something the Boomi has been working on … we’re very invested in agent security and governance/guardrails
This is definitely a work in progress as it’s not natively in MCP but I’ve seen movement in the ecosystem. Okta recently announced some work they are doing around it. We are also working on things in this area at Heroku too
I’m very interested in security around MCP servers so would be keen to learn and share learnings on this topic too!
Thread on this forum to capture collective learning and heard from some of our vendor-partners?
Please submit your feedback for the talks you attended. It’s so valuable for us and the speakers. And after all, feedback is a gift and sharing is caring! Enter your feedback for those talks here: https://etlslasvegas2025.sched.com/ https://devopsenterprise.slack.com/files/U06GCH026KT/F09GFGYPA0L/feedback.png