Veris Plus is live

Our simulation sandbox that's been powering enterprise AI agents for a year is now fully self-serve.

Reliable AI agents aren't prompted into production. They're trained. The teams shipping reliable agents today aren't writing better prompts — they're putting their agents through thousands of reps in realistic environments and learning from what breaks. That loop has been brutal to build. Most teams duct-tape it: mock services here, eval scripts there, a graveyard of YAML scenarios nobody trusts.

For the past year, only enterprise AI teams had access to Veris. Today, anyone can use it.

Think of Veris Simulation Sandbox as a video game your agent has to beat before deployment. Realistic worlds. High-fidelity NPCs similar to the users of your agent. Scenarios from happy paths to adversarial edges. A scoreboard that tells you exactly where your agent failed and why. Fail in a full realistic simulation, not in production.

You bring an agent in any framework or in no framework. You get:

• A customizable sandbox from 30+ built-in popular SaaS service simulations, 7+ modalities, and hundreds of realistic simulated users
• Scenario composing with auto-generated success criteria and behavioral graders
• Hundreds of parallel rollouts in Veris's dynamic world with simulated tools and users
• A detailed audit report that tells you what broke and how to fix it

Sign up

Plug it into your autoresearch loop. When you're ready to train models, not just iterate on the harness, the same environment is your RL rollout engine. One click.

Reliable agents are trained, not prompted. Veris Plus is your agent's training ground.

If you're building agents, build it on Veris.