3 months at langfuse.com and this keeps happening. People don’t look at numbers but just ship and it works. Focusing only on how to ship more stuff faster frees up your mind. Very few discussions like “will this work”, “will it move the needle” etc..

build something people install

Happy to get this out! If you notice any missing auto instrumented mappings from frameworks, let me know :) フィードバックをお願いします !

excited about more voice interfaces have a room for myself this week (team is traveling) and use whisper so much to get through inbox/coding

Much improved DX for all langfuse.com JS/TS users: - automated nesting via context managers and the new ‘observe’ wrapper - otel-based -> easy interop with other instrumentation - more performant across the board Huge ship Hassieb Pakzad, JS SDK is finally on par with our Py DX

after some downtime, getting back to X for the great lock-in this q4

New at @Langfuse: Natural Language Filtering for Traces. Use plain English queries to simplify trace filtering. Powered by AWS Bedrock with zero data retention.

Great ship by Hassieb Pakzad Makes running experiments against Langfuse datasets (derived from traces) and local test cases way easier. Results are available right away and can eg be used in CI

🪩The one and only State of AI 2025 is live! 🪩 It’s been a monumental 12 months for AI. Our 8th annual report is the most comprehensive it's ever been, covering what you *need* to know about research, industry, politics, safety and our new usage data. My highlight reel:

The model capability overhang is so big right now that talented product and design engineers can build on an entirely green field.

2024 evals can it count letters 🥺 can it do college stuff 🤓 are its solutions diverse 👉👈 2025 evals has it worked for 30 hours yet 🦾 has it increased gdp 📈 has it discovered novel math 🧮

Excited for this, big things planned, lots of cool things to demo from last quarter