Rivet – open-source AI Agent dev env with real-world applications
We just launched Rivet, the open-source visual AI programming environment! We built Rivet, because we were building complex AI Agent applications at Ironclad. It unlocked our abilities here, and we're excited to make available to the entire community. Backstory: A few months ago, inspired by things like LangChain and LlamaIndex, we started building an AI agent that could work with legal contracts. Unfortunately, we couldn't just use retrieval augmented generation (RAG), because a lot of contracts are basically identical (many chunks with near-identical embeddings), except for a few key…
In plain words
Rivet is an open-source visual programming environment for building AI agents. Developed at Ironclad for creating complex agent applications, it provides a graphical interface for designing and debugging AI workflows. The tool addresses challenges in building sophisticated AI systems, such as those involving contract analysis, by offering visual debugging and chain-of-thought capabilities. It is designed for developers building production AI agent applications.
written from the facts on this page · September 2026
From the sources
In the maker’s words, at launch
We just launched Rivet, the open-source visual AI programming environment! We built Rivet, because we were building complex AI Agent applications at Ironclad. It unlocked our abilities here, and we're excited to make available to the entire community. Backstory: A few months ago, inspired by things like LangChain and LlamaIndex, we started building an AI agent that could work with legal contracts. Unfortunately, we couldn't just use retrieval augmented generation (RAG), because a lot of contracts are basically identical (many chunks with near-identical embeddings), except for a few key details. So, we turned to things like ReAct and AutoGPT for inspiration. At first, things went great. We were adding agent capabilities, doing chain-of-thought prompting. But then we hit a wall. The agent became too complex. We had debugger breakpoints on almost every line of code, but we still had no idea where the agent was breaking. Every change we made destabilized something else. After two weeks of fumbling, I decided to end the project. But one of my teammates, Andy, didn't give up. The following week, he showed me v0 of Rivet. He'd used it to refactor and improve our existing agent. I was skeptical... it just seemed like a visual programming environment, and I was not a fan. But I gave it a shot, and suddenly found myself able to add new skills to the agent, debug brittle areas with ease, and update prompts with confidence. Rivet is a game-changer. And more than that, it makes building with LLMs super fun. What exactly makes it different? First, the debugger is incredible. You have to experience it to believe it. You can update a graph, and then immediately run it, and see where it succeeded or failed. Even better: you can attach Rivet as a remote debugger, and watch your agent graphs execute in your app. Second, visual programming is actually a game-changer for prompting LLMs. I don't know why exactly, but it's way easier to understand and organize your work when you have an extra dimension to work with. Finally, Rivet is built to be embedded into a larger application (TypeScript for now, but we've also found a way to run it in Python). Beyond importing Rivet as a dependency, you can also define "external functions" dynamically at run-time. It feels pretty sketchy to give a LLM a key and unfettered access to an API. With Rivet, you can give it access to a specific set of defined functions, potentially pre-scoped to the access level you want. ...Sorry that was long. If you read this whole thing, thank you! We're really excited to hear what you think! We just launched our first Rivet-based application at Ironclad, and we've been working with companies like Sourcegraph, Attentive, AssemblyAI, Bento, and Willow to make Rivet useful for others.
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 26d ago · cactuscompute.com
