BAML - A new programming language for building Agents
Agentic AI, Context engineering
As software engineers, we invent new file formats all the time.
.ipynb because python wasn't interactive enough and we wanted the tooling Jupyter Notebooks come with .tsx because React is way better for embedding HTML with TS than vanilla TS.
We're excited to share .BAML, whole suite of tooling that makes workings with models easier, not harder. And just like .ipynb/.tsx, BAML works with every other programming language, not replaces it.
We're talk about why we need very different tooling for AI develpoment, and how people are finally building robust pipelines that process 100+ page bank statements with 0 cents of error for 99%+ documents, or health care faxes direct to ERPs with 98.5% correctly automated tasks.
The goal isn't just to understand how to keep LLMs from “breaking,” but also how to design software that we can continue to maintain and develop in the world of tomorrow, where everything is stocastic.
Vaibhav Gupta is the Founder and CEO of Boundary, a Y Combinator startup developing a new programming language (BAML) that makes LLMs both easier and more efficient for developers. Vaibhav has built predictive pipelines at D. E. Shaw, Google, and Microsoft. In his free time, Vaibhav enjoys competitive table tennis, board games, and compilers.