Launching today
confused-ai
Build AI agents, teams, and workflows in TypeScript
2 followers
Build AI agents, teams, and workflows in TypeScript
2 followers
Build production-grade AI agents in TypeScript. 40+ LLM providers, 100+ built-in tools, multi-agent orchestration, RAG, guardrails, circuit breakers, HITL, budget enforcement, OTLP tracing. Open source. CrewAI / LangChain alternative.
Hey ProductHunt! 👋 I'm Raja, the maintainer of confused-ai — and I'm genuinely thrilled to share this with you all today.
The honest reason I built this: I kept running into the same wall. Every time I wanted to ship a real AI agent in
TypeScript, I had to either wrestle with Python-first frameworks or stitch together a dozen packages with no clear story for how they'd scale. I wanted something that started simple — one agent, one tool, one run — and grew with the system, not against it.
That's the core idea behind confused-ai. You start with a single agent and a model. When you need live data, you add tools. When you need continuity across turns, you add sessions. When your answers need to be grounded in real documents, you add retrieval. When one agent isn't enough, you build a team with specialists and a supervisor. Every layer is optional and additive — you never have to rip things up to move forward.
What I'm most proud of is the production side. Shipping an agent is one thing; running it safely at scale is another. confused-ai has circuit breakers, budget enforcement, PII guardrails, human-in-the-loop approval checkpoints, and full OTLP tracing built right in. You don't have to bolt these on from five different libraries.
It supports 40+ LLM providers out of the box — OpenAI, Anthropic, Gemini, Groq, Ollama, and many more — so you're never locked in. And because it's plain TypeScript with a single npm package, there's no config DSL to learn
This is fully open source under MIT. I'd love for you to try it, break it, and tell me what's missing. Every question and piece of feedback here genuinely helps shape what gets built next. Thank you so much for the support — it means everything to this project. 🙏