Prasad Ovhal·Nov 2Ego-Driven Development: I Told AI Agents Their Rivals Did BetterI tested whether rivalry prompts improve AI agents across three experiments. The early results hinted at a benefit. A larger follow-up did…A response icon1A response icon1
InGoPenAIbyRashmi·Oct 25AI Agent Evaluation: Testing Tool Calls and Task Success in ProductionBeyond LLM-as-a-Judge: A practical guide to evaluating AI agents using tool-call accuracy, trajectory analysis, task completion, safety…
Adnan Masood, PhD.·Oct 23An Illustrated Guide to AI Agents: A Book Worth Reading With a Pen in HandMy review of Maarten Grootendorst and Jay Alammar’s new book, its illustrations, and the TinyAgent codebase — with some thoughts on why a…
Indata from the trenchesbyAdel Hassen·Oct 22Agent evaluation: What to measure, how, and whenThe key to shipping agents with confidenceA response icon1A response icon1
Suresh Parimi·Oct 18Beyond the Demo: How to Engineer AI Agents You Can TrustThe architecture of execution, evidence, system intelligence, and production assurance.A response icon1A response icon1
InDatabricks for StartupsbyYuval Danino·Sep 21See it, Grade it, Query it: Running an AI Agent like a Real Production System on DatabricksObservability, evaluation, and governance for AI agents, without ever moving your data off the lakehouse
Suresh Pawar·Sep 16Evaluating RAG in Production: What We Learned Beyond “The Answer Looks Good”What we learned about evaluating retrieval quality, safe abstention, and AI behavior in production.
Praveen Naika Shivakumara·Sep 16AI Is Evaluating AI. What Actually Proves the Agent Did Its Job?How to measure agent outcomes, test the AI judge, and build an evaluation process grounded in evidence.A response icon1A response icon1
Sahil Chutani·Sep 14What I Found When I Went LookingHow I built an auditable deep research evaluation seed set from public human conversations
Jing Ge·Aug 29The Best Practices Guide to Mature AI Evaluation: Building the Evaluation Framework — Part 3…Part 1: The Golden DatasetA response icon1A response icon1
