Cut a full release of the llm-polyglot lib today, currently [email protected], all tests passing, including Anthropic function calls, streaming, super nested schema objects/arrays/etc. - Instructor PR is now ready - I will cut the v1 release tonight (!) with better type support for

[email protected] just published with support for new anthropic tools API - has updated peer dep for latest anthropic SDK - much more reliable now that it doesn't need any of the XML shenanigans - patch coming to Instructor shortly with updated tests/types

“I'm not totally sold on the idea that future AI engineers will be TypeScript engineers, but I have to say Instructor.js worked like a charm with Next.js. “ maybe not the folks ripping LLM training in 1000lines of C - but most engineers building with AI aren’t exactly 'AI

OpenAI shared a customer story for our work with Emily Oster to build Zelma. Her team painstakingly gathered and cleaned all historical US grade school test results. So to make the data accessible, we built an LLM-powered app where users can explore the data just by asking.

Full story here: openai.com/customer-stori…

npmjs.com/package/evalz I shipped this a while ago but haven't had a chance to spend a ton of time on it - I just pushed up a release with a bunch of new stuff: - Model-graded evals w/ OAI + Instructor - Accuracy Evals (Levenshtein + cosine w/ OAI embeddings) - Context Evals -

We’re building a team of creative problem-solvers to work on building new product. I’m looking for hackers, builders, and unconventional thinkers. Generalists will thrive. Passion > pedigree. If you're the kind of engineer who can dream up an idea and bring it to life