3 Apr 2026: AI Evals, Logs and Simulations with Maxim

Ricardo Updated by Ricardo

See What Your AI Agents Are Actually Saying

Your AI agents are having thousands of conversations on WhatsApp. But once those conversations happen, they disappear into a black box. Is your agent hallucinating? Drifting off-topic? Leaking personal information? You wouldn't know until someone complains.

Today we're launching a new integration with Maxim that gives you full visibility into every AI interaction in your Journeys.

Logs capture every conversation your AI agents have — what the user said, how the AI responded, which model was used, how long it took, and what it cost. Follow full multi-turn WhatsApp conversations from start to finish, filter by Journey or AI card, and sort by cost to find expensive interactions.

Evals automatically score your AI's responses for things like faithfulness, relevance, toxicity, clarity, and bias — so you don't have to manually read every conversation to catch quality issues.

Simulations allow you to test changes to your prompts before you ship them to your users, and run Evals on them, enabling you automate your testing and ship with confidence.

How to setup

Setup takes about five minutes: create a free Maxim account, grab your API key and Repository ID, and paste them into Settings → AI → Evals in Turn.io. That's it — logs start flowing immediately.

Read the full setup guide to get started.

Was this article helpful?

20 Mar 2026: Track Conversions from Click-to-WhatsApp ads

Contact