TheOneCoder

I challenge you to find a repetitive task you can do on computer that can’t be accomplished by AO. github.com/launchapp-dev/…

@jahirsheikh8 I think the thing we are forgetting is staffing numbers. They keep getting smaller, I used to get staffed on projects with way more engineers. that’s changing.

Software engineering in 2026 is absurd. Developers pay $100/month for AI tools to write half the code. Companies still pay them $10k+/month to ship. So now grown adults are making six figures… by reviewing AI output, drinking coffee, and saying “looks good to me.” And somehow we all pretend this isn’t happening. What a time to be alive.

@rezoundous I do using animus, I have a fleet of different agents working on all the little things I never get time to do.

Does anyone actually run 20 agents in parallel?

@weezerOSINT And this is exactly why we built animus local first.

Lovable has a mass data breach affecting every project created before november 2025. I made a lovable account today and was able to access another users source code, database credentials, AI chat histories, and customer data are all readable by any free account. nvidia, microsoft, uber, and spotify employees all have accounts. the bug was reported 48 days ago. its not fixed. They marked it as duplicate and left it open.

@theo Honestly less gui. I run an army of agents with no gui. I just talk to a central agent, when I need to coordinate. This is the future

Was I right?

I mean I am making more open source than before. My biggest limitation for sure is tokens. But using animus I have setup workflows that can run weekly/daily and upgrade packages, automatically troubleshoot issues, bugs and release fixes. Great for projects in maintence mode. I think if open source projects, get community sponsored agents to help with maintenance we can see certain open source projects live much longer than before.

What happens to open source when AI is writing 100% of the code? I've been thinking about this a lot. Like… the whole system was built around humans valuing the act of contribution. You learned, you struggled, you submitted a PR, you got feedback, you got better. That loop created engineers. It created community. It created ownership. If AI writes the PR, who owns it? Who learned from it? Who's gonna stay up at 2am debugging the thing they shipped because they actually care? The cool part about OSS is that no one owns it. As a consumer, you could always look under the hood, fork it, take it somewhere else. I don't think open source dies. But I genuinely don't know what it becomes... Any ideas?

@garrytan I mean honestly, why do we need a fancy ui? I got stuck at first building the ultimate “UI” for animus. Ended up settling on a CLI and MCP. The future won’t be Ui when the main way we interact with the digital world will become agentic.

The Claude Code CLI is 🔥 but the web and desktop views are uhhh

My take is this, open source = transparent. Closed source in the age of AI could mean all types of slop. If it’s supposed to run on my machine, I should be able to see what it is and what it does. And an LLm on its own, cant just copy an entire open source project from a single “prompt”. I even recently tried it before open sourcing animus. You can get something close, but unless you really think about the architecture you will miss most details. Open source can create a moat around a community, one that survives AI and competition

let’s not beat around the bush about why many apps will go closed source in the upcoming period many apps were oss in a performative way like “sure you can self host this but it’s actually super complex so good luck figuring it out hihi you will use our cloud anyway” and ppl were actually lazy to self host it, create a competitor, etc. now you can point an llm to a repo and it can self host anything in minutes or pull parts from the source code and spin them into a competitor etc without attributing to the original source and that is threatening to $$$$$$ there you go

@zuess05 True, thats been the best way to stand out with AI and demand those good hourly rates

@the_one_coder Haha domain knowledge can never be replaced💪

Genuine question. Tech companies are laying off thousands of engineers, and the ones left behind are basically just reviewing AI-generated code. But what happens in 6 months when the AI stops making mistakes? If your entire $200k job has been reduced to proofreading Claude's output, what exactly are they paying you for?

@ClementDelangue I am just waiting till running Opus level models, is just an “npm” install away on my raspberry pi 🤨

Weird how some people always target open-source in AI! First it was: “Open-source AI will destroy the world” (spoiler: it didn't and it won't) Now: “Open-source is a cybersecurity threat because of AI” Both narratives are far too simplistic. The truth is that the exact same risks exist in closed-source systems, often even more so. For example, in practice, APIs can create much bigger data and security vulnerabilities than open systems you can inspect, self-host, and secure yourself. And as with software more broadly, open-source often ends up more secure because it benefits from far more scrutiny than private internal systems. The reality is not “open vs closed.” The reality is that AI is raising cybersecurity stakes across the board, and we need to tackle that seriously together.

@buildwithshyam Focus on your moat, a new feature isn’t a moat. Data or unique models is a moat. Yes improve your user experience, but what can you offer that your competitors can’t just got to ChatGPT and railway or prompt their way to do.

Someone copies your app today… What’s your first move? A. Check what they did better B. Focus on my roadmap C. Ship something new fast D. Overthink for a bit 😅

@ForrestPKnight While I am over here struggling 😭 31 stars and I feel like hot shiz! What do yall recommend I do?

I can't prove it, but with all of these new GitHub repos getting 50k+ stars in the blink of an eye... pretty sure the vast majority of GitHub stars are fake now.

@garrytan I would really love it if you tried Animus with Gstack! 1. Define intelligent agents as markdown inside yaml. Connect them to skills/mcp or write the skills as system prompt in the markdown. 2. Define fat work as command phases 3. Kick off Animus and go enjoy life! It’s a completely open source tool! github.com/launchapp-dev/… All 6 templates were built autonomously in the last 2 weeks! templates.launchapp.dev and was developed by animus and Gstack completely autonomously

I found myself explaining this to people over and over again at YC today because I think most knowledge work will increasingly be encoded in markdown skills (fat skills) that work hand in hand with deterministic code written specifically to be called by agents (fat code)

@zuess05 For anyone wondering I just prompt the AI to pull my assigned tasks. Then spin up worktrees, merge worktrees after reviewing the code compiles and testing really quick. Gets sent to QA, they comment and leave logs on the ticket, I feed that back to my AI agents and they continue.

Right now I am working a contract by myself, honestly I am not doing much. Using Claude code I do most of the work. Before AI this type of gig at a company like this would involve way more people. Reason I am there is domain knowledge and I am good with AI. I could probably easily be replaced by something like github.com/launchapp-dev/…. But let’s not tell them that 🤫

AI will replace “90%” of Jobs. I am already getting insane results when I simulate entire companies using animus. We are already at a point where AI agents can take over, the problem is the systems and harnesses they run aren’t there yet. I tested Opus as a product owner and had it build 6 templates over the last two weeks using Animus as the orchestrator loop. can see them here templates.launchapp.dev The AI system did everything across the entire templates and when it was ready. It used the flagship to build the “store front”. It also tested everything over and over wrote E2E test scripts, exploration testing. My role was guide the vision and go enjoy life.

You may agree or not, AI will replace all software jobs by 2030. 1. Coding jobs will be gone very first. 2. Companies will slowly phase out senior developers. 3. Junior developers will be asked to take help of AI. 4. Tech interviews will be more focused on AI awareness. DSA, Leetcode will be the things of the past. 5. A company will have multiple accounts. Each account will have one human manager who will manage multiple AI agents. These AI agents would play the role of individual project managers. 6. All projects will have maximum 1 or 2 human resource. They will just be there for critical decision making. Rest all the operations will be handled by AI agents. 7. There will be AI agents for different kind of roles. Manager, BA, Architect, Sr Developer, Jr Developer, QA, DevOps Engr, DBA, etc. 8. Companies will hire more AI agents. There will be specific companies those will prepare these AI agents. 9. Initially there will be a lot of job cut. But slowly a lot of new projects will be created requiring humans. A human who knows AI and would work with AI agents. 10. Slowly this mechanism will be pushed to other industries. Banking, Manufacturing, Transport, etc. 11. Software jobs will not remain like of today. There will be no status calls. There will be no follow up meetings. But still everything gets delivered quickly, effectively, and effortlessly. 12. It will initially impact the economy very badly. But then more and more AI enabled humans will keep entering the industry. 13. We will see more n more small companies and freelancers than big scale MNCs. Big companies will either get shut down or shift to producing AI agents. 14. Companies will pay only one time for purchasing AI agents. Hence, more money gets saved. The average income of human employees will be 3x to 5x as of today. 15. This will in turn make everyone richer. Economy will flourish rapidly. Do you agree with this analysis? What do you think how will AI impact us? How should human be prepared for all this?

Honestly I feel like I am learning more, since I can actually apply my learnings on things I read. I am going a lot more from theory to application. One example I rewrote animus using animus, but this time for multi process architecture I copied how MCP works and setup everything using STDIO. It allowed me to setup a very flexible system where I can completely replace down stream processes as plugins. Or completely change language. It was fun and very educational

Now that we have AI coding assistance... 1. What isn't worth learning anymore? 2. What's still worth learning?

How much yall bet, in 3 months senior engineers will quietly switch back to Claude and then 6 months after that switch back to codex. All of them will have rate limit issues, impossible at 200$ plans. The solution in the long term is we need to build more context efficient systems. Start engineering workflows the right way. I am squeezing a lot of sessions and work out of Claude and codex cause I split things up.

SENIOR ENGINEERS ARE QUIETLY SWITCHING FROM CLAUDE CODE TO CODEX AND HERE'S THE BRUTAL BREAKDOWN a 14-year principal engineer spent ~120 hours co-developing (not vibe coding) across both tools on an 80k LOC python/typescript project. here's what he found: Claude feels like an engineer on a time crunch: > speeds toward getting things working > ignores CLAUDE.md at least once per session > leaves tasks half-done mid-migration > changes tests to match what IT thinks the goal is > almost never creates new files — just bloats existing ones Codex feels like a 5-6 year senior: > stops mid-task to rethink and refactor unprompted > never once ignored AGENTS.md > doesn't extend god classes — it factors them out > does things you hadn't thought of that are actually additive > you can fire it off and come back when it's done the raw numbers: > Claude: more done per session, more cleanup every few days > Codex: 3-4x slower, but the work is just better > Codex Pro x5 ≈ Claude Max x20 in usage caps the real difference: > Claude needs a skilled, focused driver or it goes off the rails > Codex demonstrates competence and earns autonomy his verdict: > vibe coding a weekend project? Claude wins > building enterprise software? Codex wins "Claude requires a skilled, focused driver more than Codex does" both give crap output if you don't know SWE. the tool isn't the skill.

See let me tell you, how I do it now in 2026: Step 1 - I get a max plan Step 2 - I install Animus github.com/launchapp-dev/… Step 3: I craft the dream agentic team for my project. From product owner to dev. Step 4: I draft a proper vision. Step 5: I let loose and then go swim and enjoy life!

I tried to hire a guy to vibecode apps for me. Essentially, I tried to respect what all the AI companies advertise Basically, choose an average Joe without knowledge of programming. Got a guy after 2 days, and he started coding. Got the $20 claude for him. He did the landing page within 5 minutes. I said "Amazing, lets build the app now" "Should take us at most 48 hours" After 15 minutes, the guy hit the limit "I'm not gonna wait 3 hours for the limit to reset" I have to get this guy the max $200 plan, so i purchase it He restarted work, and everything works great. App looks bad, and doesn't really work well. I will let the guy go, and do it myself Good thing, is that we are not hitting the limit anymore. I guess its a win, so much for vibe coding.

I use different models for different things across. I run opus as conductor, with codex as a co-conductor whose goal is to review all decisions by conductor. For testing I purely use sonnet to run the tests, I didn’t find any benefit to using opus. Also testing agents will only focus on their directive. So it won’t find glaring UI issues for example unless instructed to look for it. But for me my main use case is ensuring the functionality itself works. I usually log enhancements manually for visual issues. If you are curious I built an open source tool to help me setup my AI teams properly. github.com/launchapp-dev/…

@the_one_coder sounds great 👍 you run all agents with same model or different models?

If AI writes the code, and AI writes the tests, and AI reviews both ... how do you know your software actually works as expected? 🤔

@iuliatech Honestly I don’t burn out when I use github.com/launchapp-dev/…

How do you handle burnout as an indie builder? Looking for inspiring strategies 👀