mmullins.coginiti.co

Somebody make it make sense. Databricks is raising yet additional funding, beyond their last 16th round of funding, at a valuation that is north of 35x current revenue. #databs

Databricks is Raising a Strategic Round of Funding at a $188 Billion Valuation databricks.com

Two more reasons Coginiti 26.6's LLM blocks hold up: 1. Same block runs on Snowflake, Databricks, BigQuery, Redshift, Postgres, SQL Server, and with OpenAI, Claude, Gemini, or local Ollama. 2. Batching patterns keep prompts bounded, so it scales from 50 rows to 50k.

Process large datasets with LLM blocks | Coginiti Documentation docs.coginiti.co

The thing I like most about Coginiti 26.6's LLM blocks: the model's output is typed, so it's queryable immediately. JOIN it against real tables. Write tests that validate it. Cache it. Schedule it. AI output held to the same bar as every other transform in your pipeline.

Building an AI-Powered Data Pipeline with LLM Blocks | Coginiti Documentation docs.coginiti.co

Coginiti 26.6 is out and you can now call an LLM directly inside CoginitiScript. New llm block type, sits next to your SQL. Write a prompt, declare a typed schema, get structured rows back, then query them downstream like any other table. The model's a pipeline step now, not a bolt-on.

I'm concerned about the rampant Chronic Intent Attribution Syndrome (CIAS) — the tendency to interpret a model’s output as evidence of private motives, elaborate plans, and a burning desire to fool the evaluation team — running amok in AI labs.

finance ARR ≠ sales ARR ≠ marketing ARR none of them are wrong. they're all "correct" against three different definitions. now ask an AI agent for revenue and watch it pick a fourth.