Hi Product Hunt I m part of the team behind Rhesis AI, and we are so excited to announce that we are launching next week (Nov 11)!
Building enterprise AI agents breaks down when engineering, product, and domain experts work in silos. We saw a massive need for a unified workspace where teams could evaluate, observe, and optimize agents together. Rhesis AI is designed to be the dedicated collaboration layer for AI teams, turning expert knowledge into verifiable, production-ready agent performance.
Over the recent months, we ve shipped massive updates to support the entire lifecycle of your AI agent. Here is how Rhesis helps you go from requirements to production:
Team Requirements & Test Suites: Start by collecting requirements seamlessly from your team. We help you explore relevant test cases and use expert knowledge to generate agent-specific and context-specific test suites.
The Architect: Set up your testing foundation through an intuitive conversational interface. The Architect interactively builds your test setup and provides clear visual diagrams so you immediately understand your test results.
Agent Simulation: Execute your test suites dynamically. Our simulators (meet Penelope and Polyphemus!) run realistic multi-turn dialogues to (stress-)test your agents and hunt for vulnerabilities.
Expert Annotations & Review: A unified UI for domain experts to review simulated outputs and provide the nuanced, turn-by-turn feedback that only humans can give, without touching any code.
Requirements Insights: We provide a simple overview showing which requirements are met and how they were tested, giving you the insights you need to make educated decisions on releases and launches.
Real-time Observability & Monitoring: Once your agent is in production, use our OT-based tracing to monitor latency, token usage, and costs. Turn production failures directly into future test cases and start analysing their root cause.
Hi Product Hunt ๐ I'm part of the team behind Rhesis AI, and today we're launching!
Enterprise AI agents break down when engineering, product, and domain experts work in silos. Feedback lands in Slack, sheets, and meetings, and never makes it back into the agent. Rhesis AI is the collaboration layer for teams building AI agents: one workspace where the whole team tests, observes, and improves agents together, turning expert knowledge into agents you can verify.
Over the last months we've shipped a lot. Here's how Rhesis takes you from requirements to production:
๐ Requirements & test sets: Collect requirements from your team and the documents your experts already work by. Rhesis turns them into test sets built for your agent and its context.
๐๏ธ The Architect: Set up your testing foundation through a conversation. The Architect turns your PRD into requirements, metrics, and test sets, and shows you visually how they connect.
๐ต๏ธโโ๏ธ Agent simulation: Penelope, our testing agent, runs realistic multi-turn conversations against your agent. Polyphemus, our model for adversarial prompts, supplies the attacks that safety-aligned models refuse to write.
โ๏ธ Expert annotation & review: Domain experts read conversations, mark the turn where it went wrong, and say why, in their own words and without touching code.
๐ Requirement coverage: See which requirements pass, which fail, and in which conversations. You get a clear basis for deciding whether a release is ready.
๐ฌ Observability: In production, OpenTelemetry tracing shows every step your agent took, along with latency, tokens, and cost. Turn any failed conversation into a test with one click and trace it back to its root cause.
๐งช Experiments: Change one thing (model, prompt, retrieval), run it against the same test set, and ship it only if quality holds at the cost you expect.
The goal: feedback that keeps coming up becomes the test sets and metrics that run on every change.
๐ Rhesis is open source. Self-host it with Docker Compose, or use the cloud.
For the PH community: Rhesis Cloud is free forever for 3 seats, 3 endpoints, and 500 test executions a month[, no credit card required]. That's enough to take one project through the whole loop.
๐ Try it: app.rhesis.ai
โญ Code: github.com/rhesis-ai/rhesis
๐ Docs: docs.rhesis.ai
We'd love your feedback: where it breaks, what's missing, what confused you. The whole team is in the comments all day.
Huge congrats on the launch, Nikolai and team! Great to see Rhesis live on Product Hunt. I believe thats really cool to have one place where people can actually review agent runs in a browser while devs pull those exact cases via MCP, I mean is so much better than chasing feedback across Slack and spreadsheets. Looking forward to integrate when we go live as well.
We are proud to launch Rhesis AI for the community today! Our team has put a lot of love and care into the product!
Explore the Rhesis AI YouTube channel for product demos, tutorials, technical walkthroughs, and recordings of talks and community events. Learn how to test, evaluate, and improve GenAI applications with Rhesis, from getting started with the platform to exploring advanced testing workflows and agentic evaluation: https://www.youtube.com/@rhesis-ai