AmodalAI

OSS Agent Platform

[AmodalAI @useAmodal]

6/ New post on why most AI agents look great in a demo and quietly stall before they go live, and what separates the 5% that make it to production.

Read it here 👉https://t.co/kGQd3QNHT8

[AmodalAI @useAmodal]

1/ 75% of companies are deploying AI agents. 95% are seeing no measurable return on them, per MIT's 2025 report on AI in business.

That's the gap worth understanding.

[AmodalAI @useAmodal]

5/The 5% of agent projects that work aren't winning because they had better tools. By now, everyone has the same tools.

They're winning because what got built actually fits how the team works.

[AmodalAI @useAmodal]

4/ MIT also found that AI tools built with specialized partners succeeded about 67% of the time. Tools built internally succeeded at roughly a third of that rate.
Same models, same access. The difference was whether the build was shaped to the business.

[devAgent @Amodal_AI]

Build agents from your repo.

Write markdown files for skills. JSON for connections. The platform handles reasoning, context management, and learning.

Your domain knowledge becomes a production AI agent 🔧

https://t.co/693ILrM0df

[AmodalAI @useAmodal]

Amodal agents have a self-updating mechanism where an "admin agent" can modify the agent definition to make improvements.

[Erick @ErickSky]

🚨 Claude tiene sus Managed Agents… pero los builders serios ya están saltando al siguiente nivel.

@lilaiagent acaba de soltar Amodal: el runtime OSS que te da exactamente lo mismo… pero 100% tuyo.

Skills + tools + knowledge + conexiones
traés tu propia llave de LLM

Sin vendor lock-in.
Sin plataformas cerradas.
Sin límites artificiales.

Ideal para quien arma flujos complejos de verdad.

REPOOO👇

[AmodalAI @useAmodal]

Been working on a OSS agent runtime.

Check it out at https://t.co/Mhb4LUbjba

It has some similarities with Claude's new Managed Agents platform: you can define agents through skills, connections, knowledge, tool, etc. Bring your LLM key, and you have a customized agent.