Coding Agents Don't Scale Themselves. Neither Do Your Teams. — Patrick Debois, Tessl

AI Engineer · 22 min · 236 sentences · from YouTube's caption track

Each timecode opens YouTube at the start of that sentence. Line anchors (#s42) are the cue ids in the WebVTT, and every line carries its start and end seconds. All transcripts has every talk, and the whole corpus as one file.

00:01[music]
00:12Well, welcome.
00:14Um last day, I guess.
00:16That's what happens.
00:18Um I'm going to talk to you maybe not on the technical side, but more on the organizational side.
00:23So, if you're here for any technology, you can still leave if you want to.
00:31So, in 2009, um a lot of people were telling me the idea of continuous delivery was crazy.
00:38And I feel we're in kind of the same era or kind of the same thing right now with a dark factory.
00:44It will not work here.
00:46That's what I keep hearing over and over again.
00:50Um but what they're actually signaling to me, we're not ready yet.
00:54So, it's not the technology that can't make it work.
00:56It's not something they won't be able to do eventually, but they're just not set up for this.
01:04And there's been a lot of conference talks here about optimizing agents with loops and harnesses and all those pieces, and I think that's great.
01:13But eventually, we'll get there, right?
01:15It's not that this is the rocket science.
01:17And yes, we'll have to assemble this in a good way, but one day, this will kind of become commodity.
01:24Somewhere maybe even going into one of the, you know, frontier labs that just offers this as a service and will kind of make this work.
01:32Uh and that's not going to be the differentiator um for your organization.
01:38So, I'm starting from there up.
01:41Assume we're heading towards the dark factory, some kind of form of autonomous working within an organization.
01:48Um what I've seen for the people adopting this within our organization, including here where I work at Tessal, it changes dynamic of the way you collaborate around us.
02:00And for those familiar, there's like Conway's Law, like, you know, the way you organize yourselves and the tools, there is a relationship on how they interact and kind of work together on this.
02:12But today, I'm not talking about like how do you become better with your agent, but it is about how will will change your team dynamics, your platform, and your organization.
02:22So, that's what I'll take you through.
02:26Enabling the team.
02:27I assume most of you somewhere work in a team and that you're not somewhere a solopreneur
02:31[music]
02:32working.
02:33So, it kind of works different than just you with your Claude code and a team working together around that with Claude or any of the coding agents there as well.
02:46The narrative that I heard a lot is well, the developer eventually becomes more of a conductor and an orchestrator of agents.
02:56And then I think that's fair.
02:58That's been an evolution that we're on on the path where more like becoming the managers of the agent, they're kind of dealing with the agents.
03:06Now, what I've seen is that if eventually a lot of developers told me, "We didn't sign up for this.
03:12We didn't sign up for better prompting, writing better specs.
03:16We're engineers.
03:17We're technical."
03:18And that creates friction, like, is this the role that we really want to do?
03:24There was a thing that came around which maybe is more context engineering that put a first step around like, "Hey, it's not just a prompt.
03:33We'll test the prompt.
03:34We'll kind of evaluate the prompt.
03:36We'll kind of distribute the prompt and kind of optimize the prompt."
03:39So, yes, there's a little bit of engineering, but still a lot of developers kind of felt empty just working kind of with a prompt and a specification as such.
03:51What I've seen is that when we started introducing harness and loops and eventually more autonomous work within the whole organization, a new technical path opened.
04:02All of a sudden, we were helping the agent with tooling, building tooling for the agent, and that kind of reignited some of the developers who kind of felt that it wasn't for them.
04:14Now, all of a sudden, they were like, "Yes, we can do this.
04:17We have that knowledge.
04:18We're like somehow helping this even with a kind of programmatic way."
04:23So, I I think that's interesting that the identity, where we say abstraction, abstraction, abstraction, technically, all of a sudden, the craft created some new location for more engineering stuff to go to.
04:37Now, when I get the question, "Can we please help people?"
04:42And there's skeptical people, what do they do?
04:45And I always really say that these are really great people to engage in creating better context for the agent because you tell them, "Please improve.
04:57Please put all your knowledge to improve the result of the agent."
05:00And the same with the harness.
05:02So, if you have those kind of more resistant people that like complain maybe about the quality that things were produced by just the vanilla kind of coding agent, use almost that anger, use kind of that skepticism
05:16to kind of make it better.
05:20And the big mentality shift, if I would advise a a a a company right now for their developers, is kind of stop fixing the code that the agent kind of produced,
05:34but improve the system.
05:36I'm I have I'm not the only one saying this in this event, but kind of that is the difference.
05:41Like you kind of improve the system.
05:43And I think it was Swyx uh a couple of years who said it, like stop building the thing, but build the thing that builds the thing, right?
05:51So, we going on that abstraction where that is with context, with harness, with loops.
05:56And that is kind of the change that a lot of people who are still very tightly in the loop, auto completion, prompting, that they kind of need to think about elevating this to the system thinking.
06:09So, what we're really trying to do is minimize the human touches, but still with good engineering practices.
06:19And some of the narrative that comes up more often in the beginning, we're like, "Oh, great.
06:23I write code in a prompt, and then it gives a result, and we can keep going."
06:27Where we now see, well, we're kind of instructing it through prompts, but we're also instructing this like, "Please do it with tests.
06:37Please update the documentation.
06:39Please do this."
06:39All the things that we're saying to good engineers, we're now asking the agents to do.
06:45So, if you still have people who kind of yoloing their way into this, I think you should tell them, "No, stop doing this."
06:53Like, engineering practices still matter for you to maintain the system, and also for the agent to keep getting better at this.
07:03What I started seeing in some of the more advanced kind of teams is that their rituals of "Hey, we're doing a planning, and we're doing a retro in a team."
07:15That they weren't about like, "Hey, we had issues with the code."
07:19But we're saying, "We had issues with the system."
07:22So, on the retro part is like, "Hey, the agent went over and over hit this problem.
07:29Can we fix the system?"
07:31That's something you'll learn in the retro.
07:33And on the planning side, what I started seeing is that things who were that were sufficiently scoped enough were easy to pick up by agents because they were well-defined
07:46and what still was left for the humans were the things that weren't scoped out well.
07:51So, we were like a split in the planning where we said, "These things can straight go into agents, well-defined, and the harness is getting better, and this is conversational
08:00things that we need to decide as a team."
08:04And what I find important is you there's a certain kind of cycle that developers go through.
08:12Yes, they learn first about prompting, they get better, specs, context, harness loop.
08:18Also, the industry is learning like that.
08:20But, there is the lead of the team can say, "Well, stop prompting.
08:27Make the context reusable."
08:29Now, we got that.
08:30Now, we jump to the next.
08:31So, part of the team lead is putting that pace and almost that constraint and that directive in the team where it is doesn't work where you just say, "Go figure it out and do something on your own."
08:45And one of the impacts of that is that if you start producing as a team more, the people downstream, GTM, people like that, they have a hard time keeping up.
08:58Even users have a hard time keeping up.
09:00So, you need to help them also with automation.
09:02So, your harness doesn't stop at your coding.
09:05It also is extended to those people as well.
09:09And the same thing with kind of requiring uh like gathering requirements, the input might not come fast enough for your team.
09:16So, that's another kind of piece that you need to tap into that workflow as well.
09:24There's a lot of metrics that people are saying like, "Hey, is your like tokens spend and all that stuff?"
09:30I started to believe in these two metrics kind of see on how to be more productive.
09:37One is you start measuring how many human touches you still do to have the agent do the right thing.
09:46That's supposed to go down the better your harness is, the better your context is, the better your guidelines are.
09:53And on the other hand, if you're going from solo to shared system, that becomes a multiplier.
10:01You fix something once, everybody gets the benefit.
10:05This is not the multiplier from the one person becoming the 10x person, but the one change that optimized the agents has an impact on all the people.
10:16So, that is kind of the part that we're all You can start that in a team working together within your repo, sharing the context, working on a harness.
10:25But what you basically want to do is you want to scale this out.
10:28So, you come into the realm of the platform people, right?
10:32Because they're the typical shared organization working on this.
10:36Now, the platform people, they might not be paying close attention because they're like infrastructure and cloud and working on like MCP gateway and stuff like that.
10:46But there's new things like bubbling up there.
10:49They need to think about like maybe skill registries or eval systems for your context and guardrails specifically for coding agents and identities and stuff.
10:59So, they need maybe a little bit of a hand kind of growing to that role.
11:04And that kind of central role, it's hard.
11:10You need an owner to drive that program, but is it the platform team?
11:15Is it developer experience team?
11:17They don't typically own any of those pieces of the infrastructure and the other people don't really do the development.
11:24So, there's somewhere a blend, but you need to kind of make sure that there's an owner driving this centralized piece and not just within your team.
11:34Because you won't have paved roads.
11:36And that's how I see it.
11:38Reusable context across teams.
11:40Why are we all inventing how we do the authentication system?
11:44Right?
11:44This is a shared component.
11:46Let's put it in the registry.
11:48Why are you building all your harnesses?
11:50Well, if we're all using the same linters and the same security tools, that's a reusable component.
11:55So, I think that will centralize similar to the paved path for cloud into that platform registry of reuse.
12:05But, if everybody can put stuff like on the internet in a repo, it becomes a sprawl.
12:13And it becomes a thing like, well, he has a skill, he's maintaining it.
12:18That person is also has a similar skill and forked it.
12:22Now, what I do?
12:23Like, which one do I pick?
12:25So, there is a kind of thing that you say, there's an owner for this area.
12:30And they also care about making it testable.
12:33They make sure that it's modular, that other people can extend kind of the context, for example, or the harness, that it's security scanned.
12:40So, you build kind of a more centralized and the fact that it's secured and kind of maintained as something instead of just something I share around in my organization.
12:52Now, that consensus is hard.
12:54I'm not saying this is tabs versus spaces, but at times it feels like that.
12:59If you have two developer teams having to have consensus on the how the way they work, that requires a lot of communication and brokerage.
13:07So, you probably don't end up with one thing, but a catalog of three, four paved roads where they can pick off.
13:13And they can still do their own, but that's on their own budget.
13:17Right?
13:18The centralized pieces will be maintained, and that is supposed to be the easy way of adoption to go there.
13:26Now, if they do this blindly, we also want to make sure that they know what it costs.
13:33Because if we visualize the cost, they might be eager to do some optimization in there.
13:38Right?
13:39And that kind of is part of the platform team is making that visible.
13:43How much is he spending?
13:44How much is that kind of like helping?
13:46If I can reduce the number of iterations the agent has to run through, that is an optimization that I can run.
13:52But if I don't visualize that and I just see the end result, then we don't know, right?
13:57So, that is part of the platform team helping people.
14:01And so, what I'm arguing is that we should somewhere move from the solo developer to the team shared kind of context and pieces to a multiplayer system in the organization.
14:13And I think that's where the multiplication effect will happen.
14:17Right?
14:17Because you're have this flywheel of improvements that go into multiple directions.
14:25Now, one layer higher, the VP of Engineering says, "How do I enable the organization?"
14:31Right?
14:32And that is that I you know, I can predict the story in your organization.
14:37Hackathon, a lunch and learn, let's share the successes, have a shared Slack channel, have a champions program.
14:43That's all generic transformation.
14:45It could have been Agile that transformed like that.
14:47It could have been DevOps.
14:48It doesn't matter.
14:49And on the other side, we know that the strategy of just, you know, give life to something and educate people, do something, let a thousand flowers bloom, it doesn't work.
15:01So, what I'm advocating is that the kind of on the organizational is that you give the team leads and the platform that mandate to start doing that work.
15:11And it's not the solo developer piece.
15:15Now, finding people that help you externally is is mess.
15:21Yes, we have all the titles, the new job titles, AI product engineer, forward deployed engineer, you know, there was a whole talk on this, agentic engineer, AI engineer.
15:29It doesn't mean anything.
15:32You cannot judge whether what the kind of the maturity of this because nobody's really that mature.
15:40But it's a signal when you put a job posting out there that people might with the new intention will be looking there.
15:47But it's not a validation of the skills as such, right?
15:51So that is challenging for people um kind of hiring people.
15:57Now, they come to the interview and I heard stories about uh people using AI to reflect uh in their ears be response to the interview person and stuff like that.
16:10I think what what I hear from most companies is they say first step is we give them an exercise and we want them to really go nuts on the AI to solve this.
16:22You know, if they have help from AI, that's all good.
16:25That shows you kind of like how much they can kind of leverage the AI to do this.
16:32Now, after they pass this, you do a walk-through and you actually say, "Please explain me what happened.
16:38Why is this a good idea?"
16:40That's where you are testing the taste and the engineering skills on why they're doing this.
16:45First part AI, then engineering.
16:47And this third thing is how do you collaborate?
16:51Are you willing to share?
16:52Are you open or are you a solo player?
16:54That's another signal that you tap into.
16:57Right?
16:58But that fits into that whole thing of like making it shareable, making it reusable, making it engineering grade within our organization.
17:05Those are the people that you look for, not people who studied ML or AI, not people who are like experts per se at the coding.
17:13There's a blend on this.
17:15Now, you might not find a person who has all three, which is okay, but at least you know, like, hey, they're very savvy on this piece, but then for the other piece, they need mentoring
17:25and they need tutoring.
17:27But, like, don't put all the three pieces into one kind of saying like they're junior or they're senior.
17:33They have like different skills on there.
17:36Now, the VP of Engineering has to defend this and they would uh have to make the case, right?
17:46Well, we have X amount of licenses that we sold.
17:49We have faster delivery, maybe they they can promise, but hard to prove.
17:52We have quality that improved, again, hard to say.
17:56But, similar to what I said with the metrics of how effective are your agents, you can show that how much turns and how much improvement that you're making on that journey.
18:09And same thing, how much there is reuse.
18:12So, it's an easier way to kind of show metrics than comparing productivity with and without agent decoding that help you in kind of those discussions as well.
18:24And so, when people say, uh the vendors are charging completely nuts, so we're going to limit the spends, you shouldn't say like, let's limit all the spends.
18:36Your reflection should be, let's optimize the spend and help them kind of reduce that uh in a good way, where that's as simple as saying, pick the right model, educate them on the model, but also on like giving them better context and harnesses because that will make your cost go down there as well.
18:55The debate around smaller and bigger teams, yes, it's nice to have like one person who can do it all.
19:01That's the ultimate dream.
19:02They can do everything.
19:04Typically, they're paired with a complementary skill, maybe PM, design, and so on.
19:09Okay, then we need a backup if one of them is on holiday so that amounts back to three.
19:15And then maybe somebody has to care about production and tickets coming in.
19:20Could be the same people if you're really productive, but yeah, you know, you lose speed of features if you're still doing bugs and that depends a little bit on your quality.
19:28And then there's the junior you want to get on the road as well to kind of make sure they're still learning what good looks like in one of those three areas.
19:37So, I think we're still limited in the way in an organization that we're not going to each team being a solo or one or two.
19:46Yes, a lot of experience, but I think that is the thing.
19:48Now, we keep investing in actually education for that piece as well.
19:53So, one of the final things is the dark factory, which is probably a dim factory.
19:58You have to see what risk you're willing to take for what features.
20:01So, not all features will become autonomous, but you can invest more in auditing like problems, like who changed the code, verifiers that kind of check whether that code was useful and when it fails,
20:14you invest in situational awareness as well.
20:17So, there's a whole spectrum from being a micro manager to being on a autonomous approval that everything kind of is correct, but you make the decision on what your risk level is.
20:29And I think your mode is capturing the knowledge.
20:32Right?
20:32The knowledge you're putting now into skills, you're in your context, and maybe in your harness, the way you kind of restrain this, your business context.
20:41And for me, that kind of brings continuous delivery actually to continuous learning.
20:47And if you ask the question of how fast can we swap in swap out something new, that's your reactive mode.
20:55And if you can improve that ultimately, it's not about making the whole system more reliable, but can I keep it reliable while changing more of the system.
21:07I'm working on a website that kind of where I try to list some of the agent enablement patterns that I described.
21:13I couldn't list them all within this time.
21:16Tell me what you're missing.
21:18I'm trying to source social kind of stories.
21:20So, if you have a story of how things are going in your organization, please tell me and I am happy to put on a link in there as well.
21:29And if you're interested in kind of the slides, happy to share those.
21:33And I think if there's one takeaway, it's not the solo player that will win the game.
21:39It's kind of like at the different levels how we improve our organizations.
21:43Thank you very very much for listening and I hope it was useful.
21:48[applause]
22:03[music]