Simon Willison leverages the DSPy framework to evaluate and refine the core production system prompts for Datasette Agent's read-only SQL question answerer. Using DSPy agents that invoke actual tool implementations against a live in-process Datasette, and a gold-standard auto-generated dataset, the project identifies promising improvements, such as including column names in the schema listing or softening advice that causes column-name guessing.

Amazon Bedrock detects AI-generated phishing by analyzing behavioral patterns rather than grammar. Its multi-stage pipeline includes authentication checks, AI model analysis with guardrails, multi-factor risk scoring, and automated routing.

In this post, we share best practices for reliable multi-turn RL training. We cover how to build a training environment you can trust, set up an external evaluation, design a reward aligned with the end task, manage what changes once the agent runs for multiple turns, and monitor the metrics that tell you when to iterate.

Learn why coding agent bills spiral out of control — and how to trace, compare, and govern spend across Claude Code, Cursor, Copilot, and more in one place.

At the AI Engineer World's Fair, Geoffrey Litt introduced the concept 'Understand to participate,' emphasizing that developers must deeply understand code changes made by coding agents to avoid cognitive debt and maintain creative engagement.

A survey of over 1,200 IT leaders reveals that 60% of businesses already use autonomous systems, and 90% of executives say their AI rollouts exceed expectations. To turn enthusiasm into outcomes, companies must address three key questions: Are employees and governance ready? Are AI tools accessible? Do employees have the necessary capabilities? This blog explores how secure, governed AI agents can be integrated into natural workflows.

AI adoption is starting to translate into real-world returns. But as efforts accelerate, many organizations are running into the same problem: systems that are too expensive, too slow, and can’t scale. Among companies with disconnected data environments, 67% cited data storage, movement, and duplication as the largest recurring AI cost. This article explores three infrastructure considerations: delivering infrastructure at agentic speeds, streamlining data, and adopting infrastructure built for AI scale.