Tanagram Roadmap: September 2024

Tanagram is no longer a nights-and-weekends project — I’m now working on Tanagram full time, with support from Pear.

I started Tanagram to explore the idea of browsing a codebase like a database. For most of the project’s existence, I’ve been building prototypes to explore the feasibility and existing tooling for parsing code and extracting the necessary information to build such a database. Along the way, I’ve learned a lot about the problems developers run into in development flows, both from my experience at Stripe and from talking to dozens of developers at a range of different companies.

These conversations have built my conviction that a database is a valid and valuable way to browse, understand, and change code, especially across large codebases that may span many (e.g. hundreds or thousands of) repositories.

I have many ideas for applications built on top of such a database, including detailed browsers to navigate code along different dimensions, the ability to build a custom schema and precisely query code, and automating time-intensive migrations and refactorings across large codebases. In the fullness of time, I see an opportunity to build a new, modern IDE, centered around code-as-data instead of code-as-text.

There’s a lot I’ll have to figure out, including the cadence and contents of future updates. To simplify things for myself1, I’ll be consolidating updates to my Buttondown newsletter instead of posting here. If you’re interested in following along, click here to subscribe.

Let me know if Tanagram sounds interesting to you, especially if this is a problem space you’d like to work on. I’m looking for a cofounder or founding engineer to help me build — we’ll be working with parsers, compiler tooling, graph algorithms, and building polished products we can be proud of.