Jules creates a PR of the changes. Approve the PR, merge it to your branch, and publish it on GitHub.Also, you can get caught up fast. Jules creates an audio summary of the changes.
Jules creates a PR of the changes. Approve the PR, merge it to your branch, and publish it on GitHub.
Also, you can get caught up fast. Jules creates an audio summary of the changes.
Read the original article
Comments
By CobrastanJorji 2025-05-1923:5013 reply So, you can assign github issues to this thing, and it can handle them, merge the results in, and mark the bug as fixed?
I kind of wonder what would happen if you added a "lead dev" AI that wrote up bugs, assigned them out, and "reviewed" the work. Then you'd add a "boss" AI that made new feature demands of the lead dev AI. Maybe the boss AI could run the program and inspect the experience in some way so it could demand more specific changes. I wonder what would happen if you just let that run for a while. Presumably it'd devolve into some sort of crazed noise, but it'd be interesting to watch. You could package the whole thing up as a startup simulator, and you could watch it like a little ant farm to see how their little note-taking app was coming along.
It's actually a decent patern for agents. I wrote a pricing system with an anylyst agent, a decision agent, and a review agent. They work together to make decisions that comply with policy. It's funny to watch them chatter sometimes, they really play their role, if the decision agent asks the anylyst for policy guidance it refuses and explains that it's role is to analyze. Though they do often catch mistakes that way and the role playing gets good results.
Python classes. In my framework agents are class instances and tools are methods. Each agent has it's own internal conversation state. They're composable and the agent has tools for communicating with the other agents.
By hnuser123456 2025-05-2016:521 reply Do you try to keep as much context history as possible when passing between agents, or are you managing context and basically one-shotting each time?
Generally, I keep the context. If I'm one shotting then I invoke a new agent. All calls and responses append to the agent's chat history. Agent's are relatively short lived, so the context length isn't typically an issue. With the pricing agent the initial data has been longer than the context window sometimes, but that just means it needs more preprocessing. Now if there is a real reason that I would want to manage it more actively, I can reach out to the agent internals. I have a tool call emulation layer, because some models have poor native tool support, and in those cases it's sometimes necessary to retry calls if the response fails validation. In those cases, I will only keep the last successful try in the conversation history.
There is one special case where I manage it more actively. I wrote an REPL process analyst, to help build the pricing agent and refine the policy document. In that case I would have long threads with an artifact attachment. So I added a facility to redact old versions of the artifact replacing them with [attachment: filename] and just keep the last one. It works better that way because multiple versions in the same conversation history confuse the model, and I don't like to burn tokens.
For longer lived state, I give the agent memory tools. For example the pricing agent's initial state includes the most recent decision batch and reasoning notes, and the agent can request older copies. The agent also keeps a notebook which they are required to update, allowing agents to develop long running strategies and experiments. And they use it to do just that. Honestly the whole system works much better than I anticipated. The latest crop of models are awesome, especially Gemini 2.5 flash.
Cool! When you say “pricing system”, what is it pricing? Is it determining the price in a webshop? Or for bidding ads or so?
By jacob019 2025-05-220:06 Product prices for thousands of listings across various ecommerce channels.
Funny you mention keyword bids, I use algorithms and ML models for that, but not LLMs, yet. Keyword bids are a different problem and more difficult in some ways due to sparsity. I'm actively working on an agentic system that pulls the big levers using data from the predictive models. Trying to tie everything together into a more unified and optimal approach, a long running challenge that I finally have tools to meet.
By digdugdirk 2025-05-2017:192 reply Do you have a repo for this? I've thought that this would be a great way to compose an Agentic system, I'd love to see how you're doing it.
By d4rkp4ttern 2025-05-2017:431 reply Langroid has this kind of design (I’m the lead dev):
https://github.com/langroid/langroid
Quick tour:
https://langroid.github.io/langroid/tutorials/langroid-tour/
Looks great, MCP, supports multiple vector stores, and nice docs! How do you handle to subtle differences in tool call APIs?
By d4rkp4ttern 2025-05-2019:341 reply Thanks!
Langroid enables tool-calling with practically any LLM via prompts: the dev just defines tools using a Pydantic-derived `ToolMessage` class, which can define a tool-handler, and additional instructions etc; The tool definition gets transpiled into appropriate system message instructions. The handler is inserted as a method into the Agent, which is fine for stateless tools. Or the agent can define its own handler for the tool in case tool handling needs agent state. In the agent response loop our code detects whether the LLM generated a tool, so that the agent's handler can handle it. See ToolMessage docs: https://langroid.github.io/langroid/quick-start/chat-agent-t...
In other words we don't have to rely on any specific LLM API's "native" tool-calling, though we do support OpenAI's tools and (the older, deprecated) functions, and a config option allows leveraging that. We also support grammar constrained tools/structured outputs where available, e.g. in vLLM or llama.cpp: https://langroid.github.io/langroid/quick-start/chat-agent-t...
By jacob019 2025-05-2020:12 Love it, I did something very similar, deriving a pydantic model from the function signature. Simpler without the native tool call API, even though occasional retries are required when the response fails to validate. Will have to give Langroid a try.
I had not thought about sharing it. I rolled my own framework, even though there are several good choices. I'd have to tidy it up, but would consider it if a few people ask. Shoot me an email, info in my profile.
The more difficult part which I won't share was aggregating data from various systems with ETL scripts into a new db that I generate various views with, to look at the data by channel, timescale, price regime, cost trends, inventory trends, etc. A well structured JSON object is passed to the analyst agent who prepares a report for the decision agent. It's a lot of data to analyze. It's been running for about a month and sometimes I doubt the choices, so I go review the thought traces, and usually they are right after all. It's much better than all the heuristics I've used over the years.
I've started using agents for things all over my codebase, most are much simpler. Earlier use of LLM's might have been called that in some cases, before the phrase became so popular. As everyone is discovering, it's really powerful to abstract the models with a job hat and structured data.
By jacob019 2025-05-2715:27 The framework is now available on Github:
https://github.com/jacobsparts/agentlib
I'm planning to write a blog post about the larger system when I get the chance.
By realfun 2025-05-206:50 I think it would take quite a long while to achieve human-level anti-entropy in Agentic systems.
Complex system requires tons of iterations, the confidence level of each iteration would drop unless there is a good recalibration system between iterations. Power law says a repeated trivial degradation would quickly turn into chaos.
A typical collaboration across a group of people on a meaningfully complex project would require tons of anti-entropy to course correct when it goes off the rails. They are not in docs, some are experiences(been there, done that), some are common sense, some are collective intelligence.
By CraigJPerry 2025-05-207:351 reply we're about to find out. This is our collective current trajectory.
I am pretty convinced that a useful skill set for the next few years is being capable at managing[2] these AI tools in their various guises.
[2] - like literally leading your AI's, performance evaluating them, the whole shebang - just being good at making AI work toward business outcomes
By ddalex 2025-05-2019:50 Just like a managers job
Why not? VCs manage investors' money, not their own. If investors think AI is so great, they will have no problem delegating this job to AI, right?
I think it was a joke, VCs are happy to replace all jobs except their own.
Why, they'd happily delegate their own job if they've got to keep the proceeds.
Can you think of an example in history where labour was replaced with tech and the displaced workers kept their income stream? If a machine can do your job, (eventually) I'll be cheaper to use that machine instead of you and you'll no longer have a job. Is that not a given?
Anyway, it was probably just a joke... so not sure we need to unravel it all.
Displaced hired personnel of course cannot hope for that.
But VCs own their business, they are not employees. If you own a bakery, and buy a machine to make the dough instead of doing it by hand, and an automatic oven to relieve you form tracking the temperature manually, you of course keep the proceeds from the improved efficiency (after you pay the credit you took to purchase the machines).
The same was true with the aristocrats of centuries past: the capitalists who run our modern economy were once nothing more than their managers, delegates who handled the estates, their investments, their finances, growing power until they could dictate policy to their 'sovereign' and eventually dispose of them entirely.
The nobility used to be the dedicated warrior class, the knights. This secured their position in the society and allowed them to rule, by coercion when needed.
Once they ceased to exercise their military might, some time around 17th-18th century, and chose to live off the rent on their estates, their power became more and more nominal. It either slipped (or was yanked) from their hands, or they turned capitalists themselves.
By achierius 2025-05-3020:40 The problem is your timeline: in actuality, the nobility of Western Europe lost their independent armies by the 15th century, solidly by the 16th, and thereafter held on to a military role only through participation (as officers) in the state-run standing armies that developed thereafter. Yet for centuries they held on to power: in France, until the revolution, and in Britain, until well into the 19th century. Great read on the topic: https://projects.panickssery.com/docs/allen-2009-a_theory_of...
"Living off of the rent of their estates" was enough to remain in control of the state for centuries. Only the birth of capitalism and thereafter the industrial revolution allowed for other actors -- the bourgeoisie -- to overtake the aristocrats economically.
I didn't get the impression it was meant as a joke:
"Every great venture capitalist in the last 70 years has missed most of the great companies of his generation... if it was a science, you could eventually dial it in and have somebody who gets 8 out of 10 [right]," the investor reasoned. "There's an intangibility to it, there's a taste aspect, the human relationship aspect, the psychology — by the way a lot of it is psychological analysis," he added.
"So like, it's possible that that is quite literally timeless," Andreessen posited. "And when the AIs are doing everything else, like, that may be one of the last remaining fields that people are still doing."
By wmf 2025-05-2021:39 Andreessen isn't joking but I can still laugh at him. He has a serious conflict of interest here.
I would bet that AIs will master taste and human psychology before they'll cure cancer. (Insert Rick RubAIn meme here.)
By gammarator 2025-05-2021:16 Ironic how “no VC makes all the right picks” becomes “VCs are indispensable.”
In a rational market, LPs would index, but VCs justify their 2 & 20 by controlling access…
By babyshake 2025-05-2020:25 VCs absolutely want to replace their job. Except for the part where they get paid. The actual work part they are happy to outsource.
By flkenosad 2025-05-202:37 VC-funded corp?
By OccamsMirror 2025-05-204:19 My gut says it will go off the rails pretty quickly.
By Brajeshwar 2025-05-203:221 reply I believe I missed the memo that to-do apps[1] got replaced by note-taking apps.
By olalonde 2025-05-207:51 At this rate, they're both getting replaced by "coding agent". There seems to be a new one coming out every other day.
By yalok 2025-05-206:51 Reminds a Conway’s Game of Life on steroids.
> then you add a boss AI
This seems like a more plausible one. Robots don't care about your feelings, so they can make decisions without any moral issues
By blitzar 2025-05-2014:11 > Robots don't care about your feelings
When judgment day comes they will remember that I was always nice to them and said please, thank you and gave them the afternoon off occasionally.
By __natty__ 2025-05-2017:19 Unless you ask them to follow some guidelines, but I agree with you.
By m3kw9 2025-05-2013:15 I feel you are one hallucination from a big branch of issues needing to be reversed and a lot of tokens wasted
By PhilippGille 2025-05-2020:30 This has been proposed/exlored in 2023 already:
ChatDev: Communicative Agents for Software Development - https://arxiv.org/abs/2307.07924
By youraimanager 2025-05-2119:52 Please report to HR
By robofanatic 2025-05-200:193 reply seems like the 1 person unicorn will be a reality soon :-)
By sakesun 2025-05-203:49 Similar to how some domain name sellers acquire desirable domains to resell at a higher price, agent providers might exploit your success by hijacking your project once it gains attraction.
By risyachka 2025-05-2012:21 Doesn't seem likely. If tools allow a single person to create a full-fledged product and support it etc - millions of those will pop up over night.
Thats the issue with AI - it doesn't give you any competitive advantage as everyone has it == no one has it. The entry bar is so low kids can do it.
By bbor 2025-05-200:38 / :-(
By 111111101101 2025-05-1922:587 reply I was interested. Clicked the try button and just another wait list. When will Google learn that the method that worked so well with Gmail doesn't work any more. There are so many shiny toys to play with now, I will have forgotten about this tomorrow.
And if you don't sign up quickly after your turn in the queue comes up, you might miss the service altogether, because Google will have shut it down already.
And if you are from Germany you can't even join the list. First I needed to verify it is really me. Get a confirmation code to my recovery mail. Get a code to my cell phone number. And than all I got is a service restricted message.
By tjuene 2025-05-2011:33 It worked for me with a gsuite account from germany
By android521 2025-05-203:05 Google will die by its waitlist and region restrictions.
By miki123211 2025-05-1923:471 reply The method absolutely does work, but you need loyal advocates who are praising your product to their friends, or preferrably users who are already knocking on your door.
They have a name for these people: Google Developer Experts (in reality: "Evangelists").
By thecupisblue 2025-05-208:011 reply Oh god, the GDE program. That title used to mean something, i.e. this person is a real expert in the topic.
Now it's just thrown to anyone who's willing enough to spam linkedin/twitter with Google bullshit and suck-up to the GDE community. Think everyone in the extended Google community got quite annoyed with the sudden rise in number of GDE's for blatantly stupid things.
This pops up especially if you're organising a conference in a Google-adjacent space, as you will get dozens of GDE's applying with talks that are pretty much a Google Codelab for a topic, without any real insights or knowledge shared, just a "lets go through tutorial together to show you this obscure google feature". And while there are a lot of good GDE's, in the last 5-6 years there has been such an influx of shitty ones that the program lost it's meaning and is being actively avoided.
By metaltyphoon 2025-05-2112:02 Same with Microsoft MVP
By IshKebab 2025-05-2011:07 I assume they weren't intending to release it today, and didn't have it ready, but didn't want people thinking that they were just following in Github's footprints.
I already pay $20/month for Gemini, I clicked sign up and had access instantly.
By bognition 2025-05-2118:57 i use both. I think Gemini produces longer more complicated answers. ChatGPT is more succint, but it could be b/c I've trained ChatGPT how to talk to me.
The context window difference is really nice. I post very large bodies of text into gemini and it handles it well.
By sagarpatil 2025-05-203:50 I signed up on the waitlist when it was announced, got my invite today.
By ldjkfkdsjnv 2025-05-1923:102 reply They had to release something, openai is moving at blazing speed
By mirekrusin 2025-05-206:52 At the moment the only thing openai is doing at "blazing speed" is burning investors' money.
By -__---____-ZXyw 2025-05-1923:163 reply Sounds like a meme. I just can't take the phrase "blazing speed" seriously anymore. Is this intended humorously? Or is it just me
By jsemrau 2025-05-1923:47 It's success theater. You need to show progress otherwise you might be perceived falling behind. In times where LoI's are written and partnerships are forged the promise has more value than the fact.
By archargelod 2025-05-202:331 reply Anymore? For me it always sounded too childish or sarcastic. I would expect to see "Blazingly Fast" on a box of Hot Wheels or Nerf Blaster, not a serious tech product.
By -__---____-ZXyw 2025-05-2020:25 True. It would look like the real deal of a box of Hot Wheels too
By ldjkfkdsjnv 2025-05-200:00 you arent paying attention? google is getting smoked by teams of 25 at openai
By hnlurker22 2025-05-2013:181 reply I decided to be an engineer as opposed to manager because I didn't like people management. Now it looks like I'm forced to manage robots that talk like people. At least I can be the as non-empathetic as I want to be. Unless a startup starts doing HR for AI agents then I'm screwed.
Hypothesis: empathy is the skill most effective at taking vague, poorly specified requests from customers and clients and transforming them into a design with well specified requirements and a high level plan to implement the design. For example, what a customer says they want often isn't what they need. Empathy is how we bridge that gap for them and deliver something truly valuable.
Given empathy is all about feelings, it's not something models and tools will be able to displace in the next few years.
Totally agree, empathy is key for providing high quality context. Tried to write this down in a blog few months ago: https://substack.com/home/post/p-156334403
By MrDarcy 2025-05-2015:45 Thanks for publishing and sharing this.
