We wanted to build with knowledge graphs. Accidentally built the plumbing. Meet yggy.
Hey makers, Carlos here, co-founder of yggy 👋
Ever started building an AI assistant and somehow ended up designing schemas, matching entities and maintaining an extraction pipeline? Congratulations, your weekend project now has a data engineering department. It’s you.
That’s roughly how yggy happened. We wanted to build graph-based products, but kept spending our time preparing and maintaining the knowledge underneath them.
Obsidian gives you a way to connect your notes. Graphify maps your codebase for coding assistants. With yggy, we’re tackling the knowledge spread across your documents, research, project notes and connected systems, and making it usable inside whatever you’re building.
Bring your data. yggy proposes its underlying ontology, extracts facts and relationships, reconciles duplicate entities and flags contradictions. Every fact keeps its source. Uncertain matches go to review, and merges can be undone.
You can explore the graph visually, query it through the API and SDKs, or connect your agent through MCP. We run the pipeline so you can spend your time on the application.
For a solo developer, that could mean:
An assistant that connects customer feedback to requirements and product decisions.
A research tool that links findings, methods and conflicting results across papers.
A shared memory that different agents can query without each maintaining its own version of events.
The useful part is that your corrections become reusable knowledge. Resolve an ambiguity once, and the applications using that graph can work from the same decision.
We’ve made yggy available self-serve, including a free plan. Bring a messy project and see how far you get before you need to touch a schema.
Replies
Max here, the other co-founder of yggy 😄
Honestly, this is the moment I’ve been looking forward to most - handing yggy over to the people who actually build stuff and seeing what they do with it.
So here’s my challenge for the community: grab that one messy project you keep putting off. The folder full of notes, papers, specs, half-finished ideas, whatever. Throw it into yggy on the free plan, plug in your agent via MCP or go straight through the API, and see how far you get. Then come back and tell us what worked, what broke and what you built. I’m genuinely excited to see what comes out of this. Let’s go!
P.S.: Personal wish list item: I want to see the first Jarvis built on yggy. If you’ve ever wanted your own Jarvis that builds its own context, one that doesn’t just link your notes or map your code, but comes up with the ontology, pulls out facts and relationships and keeps everything connected across all your stuff, this is your sign.