9 stories
Most companies cannot list the AI agents actually running. Why policies fail without an AI agent inventory, the five questions it must answer, and how to build one.
LangWatch red-teamed Langy, its own AI agent, across 15 attack categories and up to 20 turns each. Real transcripts, 14 bugs fixed, one issue still open.
Cost per successful task = run cost ÷ success rate. Learn how to calculate it for your AI agents and switch to cheaper models before flat-rate plans end.
A practical AI agent red teaming guide: attacker goals, the Crescendo multi-turn strategy, OWASP LLM Top 10 mapping, and how to test agents before shipping.
Most AI agent skills ship untested. See how LangWatch Scenario simulates full multi-turn conversations with a user simulator and judge to catch real failures.
Non-technical leaders now write agent simulations in plain language with LangWatch and Claude, testing AI agents before a single customer sees them.
Add evaluations to a multimodal AI agent with LangWatch Skills, scaffolding experiments, tool-usage checks, and Scenario simulations in about 30 minutes.
LangWatch Skills let your coding agent instrument tracing, observe production performance, and run tests automatically, no manual setup needed.
The LangWatch MCP server brings agent evaluation into Claude, Cursor, and Copilot, so you version prompts, generate scenarios, and benchmark models in flow.
Showing 9 of 9 stories
You're all caught up.
Get the top AI news and tools straight to your inbox every week.
Weekly · AI news from LangWatch Blog
Join 15,000+ AI enthusiasts
