"In this post, I’ll cover a third, not-so-obvious approach: building ways for the agent to validate more of its own work before a human has to step in. "<p>this has been an obvious thing to do since at least January (since Geoffrey Huntley published "everything is a ralph loop"), and this is how I've been working: build enough orchestration tooling to be able to automate everything: development container bringup, building it, running the unit tests, doing integration testing, and using the software as eventually an end user. then to iterate set performance goals on an already solid basis so the automated agent ("gym") can go and iterate autonomously, and let you know when it's "done".<p>I understand this probably does not work if you're on some subscription and not using the API (tokens burn fast), but this has been extremely productive for me.
You can get really far with the 20x Claude Code and Codex plans. They are many orders of magnitude cheaper than api calls.
This is where most of my productivity gains have come, I have a special harness I move from project to project now that does my testing orchestration, lots of my work day is setting up a prompt or two early and just letting them loop till they return evidence that the feature is working having gone through the big QA loop.<p>I've slowly been optimizing for token use through the stack and Claude ends up making very tight for loops for most of the process and keeping token count even lower. It's been nice. A lot of my toil at work is just gone.
What license do you use then?
Isn't this a bit of an incorrect usage of the term "backpressure"?<p>OP quoted the correct definition right at the start:<p>> <i>In systems engineering, backpressure is the mechanism by which a downstream component signals upstream that it can't accept more work</i><p>(the "downstream component" being the human reviewer in this case)<p>But the measures they propose don't actually do that. They are more like fixed throttle elements which would slow down the rate of submissions of an agent and weed out some low-quality submissions before hitting "downstream".<p>I'm missing the connection to the actual capacity (or will) that the human developers have to review the submissions.
Author here. Well noted. I do think backpressure might not be the ideal analogy/term.<p>It comes from previous posts I’ve come across, but I haven’t considered exactly what you mentioned. That’s on me.
It is an incorrect use of what was already a flawed metaphor. Pressure is isotropic. Directed pressure makes no sense, like all other fluid analogies in unrelated fields of engineering.
Wait so cross ventilation, where a breeze will flow through a house if windows are open on opposite sides at a much greater rate than if windows are only open on the upwind side… isn’t really a thing?
I took the analogy to be about the location of the pressure and not the direction. If you allow pressure to build on the input pipe when you can't accept more, the component that is upstream in the flow is able to observe that and respond. Maybe the difference is I envisioned a series of pipes and not a single one.
The act of "making pressure" means applying a force and is completely directional.
A very long post about a simple and very obvious idea with many different implementations.<p>The three main problems are 1) API usage is deadly expensive 2) Claude is about to make all automation very expensive 3) all the flows where a model has the initiative are strictly biased towards unwarranted stops (checkpointing).<p>Also, I won't call that "backpressure", there is no producer-consumer disbalance or something similar. From what I can see, the author just proposes a structured feedback loop. That's a discussion about organizational principles for system which consist of multiple unreliable but very complex components and this "backpressure" is just one of the aspects. Personally I find the viable system model framework productive as both a mental model and literal implementation guideline.<p>Lesser problem is that agent SDKs are bad and building a custom harness is hard.
> all the flows where a model has the initiative are strictly biased towards unwarranted stops<p>Can you elaborate on what you think causes such a bias? My experience is that Qwen3.6, Claude Sonnet 4.6 and Opus 4.6/4.7 will work as far as they can given direction and a way to test their work. My so-far limited experience with Opus 4.8 is that it does stop somewhat earlier for feedback, but in places where I am glad it is checking assumptions or where I agree with it identifying a change in scope (for example, where the following work deserves a separate commit or merge request). I would call those justified stops rather than unwarranted.
<a href="https://en.wikipedia.org/wiki/Back_pressure" rel="nofollow">https://en.wikipedia.org/wiki/Back_pressure</a>
The problem is not "backpressure", that's just one of the tools and there are different approaches with the same effect.<p>You can't express orchestration in terms of "backpressure" only, I think.<p>Implement-Review-Repeat loop does not involve backpressure in the strict meaning of the term.
> 2) Claude is about to make all automation very expensive<p>Wait, what happened here??
This what hooks[1] are for, except hooks allow specifying criteria in certain conditions (like the agent believing it’s done and ready to hand back to the user) in a manner that the agent won’t just forget about once it’s a few turns deep, and doesn’t require triggering a whole other LLM instance to read some plain text instructions while you hope it interprets them correctly.<p>It absolutely makes sense to have a system in place that allows the code generated by an LLM to be automatically validated but there’s no need to resort to a non-deterministic system for these sort of deterministic pass/fail conditions.<p>[1] <a href="https://code.claude.com/docs/en/hooks" rel="nofollow">https://code.claude.com/docs/en/hooks</a>
I was thinking the exact same thing. There are multiple places to implement hooks (git hooks, Claude hooks, etc.).<p>One thing I've been wondering about is how to reliably protect specific portions of the system from unexpected/unnecessary change (for example, a failing test that Claude decides to comment out or rewrite to get it to pass). My only thought for this was to automatically revert test changes during specific portions of the implementation, but that feels overly rigid and potentially prevents things like refactoring code.
I always use a standard workflow and it has never been a problem.<p>- Define the task and the goal, write a short spec document (markdown is fine)<p>- Point the agent at it in plan mode and have it write the plan to disk with phases. Iterate on its plan if necessary here and now.<p>- Have each agent tackle a phase and have it update it as a living document (switch models if some phases are more difficult than others)<p>- Clear and repeat until done<p>I've never had to overcomplicate this and it's worked both on enterprise-scale projects and personal projects. I am not sure what I'm missing - if anything.
> It should also reduce the number of low-quality PRs your teammates have to review for details the agent should have caught itself.<p>Oh boy.
Care to elaborate?
That quote shows an utter disregard for basic human decency.<p>It is the responsibility of the person running the coding agent to make sure the resulting PRs are high quality. Putting that on your team mates, or worse, random open source project maintainers on the internet, is the definition of an extractive contribution.
They are probably reacting to the laughable idea that by making PRs 20% better (or whatever), devs will continue to review the code with sufficient rigor to catch even the bugs they're supposedly now preventing. Assuming such rigor was ever present in their work!<p>Put another way, who are they supposed to hire to tell these low quality PRs apart from the high quality ones? Who even knows how to do something like that?!
This seems to be the coding agents 101: build a strong feedback loop. Am I missing something?
No, there is nothing new in this article. This methodology is already adopted - and further optimized - by the sw community.
Yeah I don’t really see the backpressure analogy here - it implies that the agent is constantly producing new stuff, which isn’t really possible since the solution is very detailed specs/goals.
Looks like plenty of recent prior art on this:<p><a href="https://pura.xyz" rel="nofollow">https://pura.xyz</a><p><a href="https://github.com/puraxyz/puraxyz/blob/main/docs/paper/main.pdf" rel="nofollow">https://github.com/puraxyz/puraxyz/blob/main/docs/paper/main...</a>
Interesting ideas for generalizing goals to reduce human labor in human <—> agent interactions. That said, maybe it is better to set up customized skills and infrastructure for large projects? At our early stage of trying to capture value of agentic systems, the good ideas in this article might be premature optimization.
If the systems invariants are well defined, and a suite of conformance + requirements tests (ensuring invariance is respected) are defined, wouldn't this be a broad - _'base case'_ - approach in general?
Everyone looking into this and other verification should be moving away from long prompts and complex skills, and looking into hooks.<p>If you put all these checks in your stop hook and your git commit hook, your repo docs can tell your agent that checks will run automatically when it stops work, and it should fix any problems found.<p>It’s wonderful to reintroduce determinism at the QA end of your process. I find it very calming to know the agent can’t skip or forget to check its work because with hooks the checks are run by the harness.
I think pi-subagents (which can form arbitrarily long chains of subagents, with up to 8 in parallel) and Claude Code‘s new workflows feature, are quite convenient abstractions that can be setup quickly.
If that's third then I have fourth. Self plug obviously, but figured that I'd like something between smart autocomplete and an agent -
an autocomplete that has wider context.<p>Called it rik, and it's on GitHub if anyone's interested checking it.<p><a href="https://github.com/exlee/rik" rel="nofollow">https://github.com/exlee/rik</a>
Oh this is 101. Anyone not doing this? If not do it now!
I’m willing to be wrong but this industry-wide emphasis on AI creative/coding workflows seems way over-engineered.<p>Ime successful creative execution looks like micro-iterations where each output informs the next creative move.<p>I can build something incredibly fast from essentially caveman grunt instructions through an LLM harness, iterating as I go.<p>Optimizing for feeding a huge plan to an agent sounds to me like a net waste of time. And looking over the shoulder of industry peers trying to do this, I don’t see their outputs or throughput some remarkable improvement over what I can produce with minimal fanfare usage.
Yeah it's wild watching so many people decide waterfall is great all of a sudden.
Every large project in the coming back to waterfall. While the problems are certainly known and it was ultimately developed as a straw man, everything else ends up working worse. That said, you shouldn't be thinking pure waterfall as it's drawn up as a strawman, but rather a waterfall variation with feedback loops. But in the end, in very, very many cases, you have to know an end date in order to get things done because so many other things depend on you being done at the same time. If something is going to get done sooner you can't use it anyway without all the other pieces.
Never mind stumbling into proper engineering principles like having documented, testable requirements specifications.
LLMs are too flaky for high quality code. On tougher problems it's very common for an LLM to contradict itself and run in circles. It simply doesn't know what the right thing is, but on each turn it is super confident to do the right thing.<p>Maybe I've chosen hardmode to learn C with LLM assistance, plus my pet project turned out to be a bit less trivial then anticipated. But I know that I have to think three times about my choices how to deal with C problems and seeing how a LLM struggles to give reasonable answers is a a huge red flag and forces me to think about it a fourth time.<p>Doing all this with a fast autonomous workflow with just little user guidance is asking for trouble.
It’s not just you. My last dumb pilot program making a Pocket clone in Python also got stuck in a loop regularly, which should be its strong suit...<p>I suspect that the “right” way to use LMs in coding, including accounting for focus, control, and costs is not a settled debate. We probably haven’t even seen the best ideas yet. But I’m really dislike the maximalist approach.
100% agree. Took me ages of working with the agents to circle back around this, which was the best way to get work done before AI automation anyways.
For me it's usually that I start with a single agent, but then I won't have anything to do while it is churning and I have other ideas/features that keep building up that I want to do, so I need to scale, and while I'm scaling I need to start to have those workflows, so eventually I end up with many agents, most which are autonomous working on their own worktrees, but I will have a specific agent that I will talk to more iteratively.<p>So e.g. I may have 1 agent that I ask and iterate on with directly, and 9 agents that work separately on their own.<p>I will utilize this 1 agent on features I care most about and want to guide and iterate on in as much detail as possible.
I suspect that letting agents spin away unattended for long stretches of time will become less and less popular as more and more companies blow their token budgets and start requiring some answers to difficult questions before agreeing to further loose the purse strings.
The trick is to have 5 of those huge plans running in parallel.
I agree. I have gotten an incredible amount of work done iterating with 5-30 minute long agent tasks. But it requires I stay engaged, and not go chill on the beach, which I guess is a lot of agentmaxxers’ goal.
you do not need micro iterations. you can set macro goals and let the agent/LLM/model whatever you want to call it figure it out.<p>it works.
interesting idea, unfortunately programming the structure is equivalent (P=NP) to just programming itself. same as TDD.<p>as usual, the tool isnt really doing whats listed on its label.<p>however, people are different so this might improve someones capability to deploy LLMs. might even provide better evidence where actual brain power is needed.
[flagged]
[dead]
[dead]