Paweł T

AI isn’t transforming the labor market. It’s dismantling it. Industry by industry. Job by job. And nobody is warning you in time. Read full article: @p.targosinski/your-job-wont-be-automated-it-will-be-deleted-984f2c40f134" target="_blank" rel="nofollow noopener">medium.com/@p.targosinski… #ai #futureofwork #automatuon #jobs

@icanvardar <3

open source will always be the winner

@GoogleAIStudio Wove - Open-source AI coding agent. Desktop app, bring your own model. Writes code, browses the web, verifies its work. Apache 2.0.

What are you vibe coding this weekend?

@suni_code Hi! Please upvote my opensource AI coding app github.com/mits-pl/wove

X is cool. But it’s 100x better when you connect with people who actually build. If you're into tech, AI, or coding, say hi 👋

Knowing what to build. AI writes code fast but it has no idea if the thing should exist. 15 years of shipping software taught me that the hardest part was never the code. It was figuring out which problem actually matters. AI makes the gap between "good taste" and "no taste" even wider.

can I ask a dumb question… if everyone uses AI… what makes someone better than others

This is the core problem. LLMs write decent code in small steps. They write garbage in big batches. Then you pay again to fix the garbage. What if the tool itself forced small steps? Read before edit. Verify before “done”. Break big tasks into isolated sub-tasks automatically. And let you swap models, so quick fixes go to a cheaper model instead of Opus. 20x less cost for 80% of tasks. Or run a local LLM on your own machine and pay nothing. That’s what I’m building OpenSource

I'm not joking, I had Claude write a big batch of code last night. I am troubleshooting rn. I asked it to review, It said this is trash code and it needs completely reworked. We are spending credits to run in circles.

@hiarun02 MiniMax M2.7. Nobody talks about it but it's scoring near Opus levels on coding benchmarks at a fraction of the cost. Dark horse of 2026.

What’s your current favorite LLM ?

@_Chandan_17 Wove + Claude + MiniMax M2.7 for coding. Built my own opensource tool because I kept hitting limits on everything else. Now I just use my API key and don't think about usage bars.

As a solo founder, which AI tool do you actually use daily? - Cursor - Claude - Codex

@AfiaDimple_ :)

If you need verified accounts then camp here.

@RiazBaby hi

Who is active Say "hii We boost U

That's exactly why we went open source and local-first. When the agent runs on your machine with your API key and you can read every line of the harness code, the liability chain is clear: your tool, your config, your responsibility. The black box SaaS agents are the ones with the liability gap. Nobody can audit what Claude Code does internally. With Wove you can trace every tool call, every decision, every guardrail that fired or didn't.

@tlewap @thunkoid @X Built agent workflows for three clients this quarter. The liability question always kills adoption faster than any technical issue - who signs off when the agent screws up a $50k decision?

Hello @X Could you connect me to a bunch of indie hackers? Also, solo founders, public builders, script writers, product owners, marketers, software developers, AI enjoyers, and designers would be great. I’m building in public, learning everyday, and enjoying the tech scene

@mscode07 github.com/mits-pl/wove

Drop your project URL Let’s drive some traffic Curious to know what you all are building 👇🏼

@lunarfq how? :)

If you reply to this post. You will gain 500 followers.

Warp was the direct inspiration. I wanted the same experience but open source and with any LLM, not locked to one provider. Wave Terminal was the perfect base already had the terminal engine, SSH, file editor. I added the AI agent layer on top: tool system, browser vision, sub-tasks, guardrails. Right now I run it daily with MiniMax M2.7 and I'm currently running Terminal-Bench 2.0 (the standard benchmark for coding agents). First pass scored 45.1% if it holds across all 5 required runs, it'll be the best MiniMax result on the leaderboard. Current best MiniMax score there is 42.7%. Claude Code with Opus 4.6 sits at 58%. The whole point: you bring your own API key, pick any model, and the harness does the rest.

@tlewap @X This is comprehensive! What was the inspiration to make it? It I understand correctly, it’s in the same space as the warp terminal?

@PaulLarionov Hi! github.com/mits-pl/wove

Hey founders ! Looking to connect with people building in: 🍽️ SaaS 🚀 Tech 📲 Automation 🧠 AI tools 📱 Apps 🔥 Web APP 💻 Devs Drop what you're working on👇

Update on Wove vs Terminal-Bench 2.0. Run 1: 27% (GitHub Actions, first attempt) Run 2: 45% (dedicated server, prompt tweaks) Run 3: 62.5% after 40 trials (harness improvements) Same model every time. MiniMax M2.7. Claude Code + Opus 4.6 ($200/month): 58% Three runs. Three days. Each time I improved the harness, not the model. Still running. 405 trials to go. @alexgshaw @MiniMax_AI #BuildInPublic #TerminalBench #OpenSource

@1Umairshaikh Running coding benchmarks on Wove, my open-source AI terminal. Testing how MiniMax holds up against Claude and Gemini on real tasks. Early results are interesting.

What are you building this weekend? Drop your project URL Let’s drive some traffic

@Ever_Featured Wove. Open-source AI coding agent. Desktop app, bring your own LLM, no subscriptions. Currently benchmarking MiniMax against Claude and Gemini for real coding tasks. github.com/mits-pl/wove/

Hello Founders You built something valuable! Now it needs visibility We help products get discovered beyond the launch hype Drop your product 👇

@harigovindcorp github.com/mits-pl/wove OpenSource

Hey founders 👋 Looking to connect with people building in: 🍽️ SaaS 🚀 Tech 📲 Automation 🧠 AI tools 📱 Product Development 🔥 Web APP 💻 Devs Drop what you're working on👇

@delveroin github.com/mits-pl/wove OpenSource Ai terminal

DROP YOUR SAAS!! Let’s send some traffic there!!!!