Databricks' KARL agent uses reinforcement learning to generalize across six enterprise search behaviors — the problem that breaks most RAG pipelines.

Many Caribbean professionals search for “remote jobs” but leave the location filter set to “United States.” They then apply to roles requiring US work authorisation — and are automatically rejected.

Databricks has released KARL, an RL-trained RAG agent that it says handles all six enterprise search categories at 33% lower cost than frontier models.

Cove Street Capital analyzes the AI market mania and shifting software valuations. Read the full analysis for more details.

I’ve been fortunate to invest in several AI funding rounds—from pre-seed to Series B to F—and to see up close how billions have flowed into ...