01 · Environments
A working copy of a business system.
A real database, tools served over MCP, an operating manual with rule IDs, tasks and a grader. Every episode starts from a seeded state.
Environments →Product
One engine for training and testing AI agents.
AutoStack sells environments for training and testing AI agents, delivered through engagements. Teams building agents start with a feasibility week (USD 5,000, credited toward a USD 20,000 evaluation) to test one workflow on a mirror of their agent's tools. AI labs get private environments built to a written brief, priced by quote. All data is synthetic.
How the product fits together
02 · Test
What the agent did, not what it said.
Run your agent on a mirror of its own tools. The grader reads the audit log, including attempts the system rejected.
Test →03 · Train
Rewards you can verify and rerun.
The grader turns each episode into a reward from state checks. Labs train on it; do-nothing and shortcut agents score zero.
Train →04 · Integrations
Plugs in where your agent already calls tools.
MCP today, and a REST mirror of your tools in an engagement. If your agent can point at a different endpoint, it runs unchanged.
Integrations →Who it's for
Who it's for
labs · AI labs and RL teams
AI labs and RL teams
Private environments with verifiable rewards, built to a written brief.
builders · teams building agents
Teams building agents
Test one workflow on a mirror of your tools before your customer does.
insurers · claims AI oversight
Insurers overseeing claims AI
An independent evaluation of a vendor's agent, run against your rules.
Engagement
Start with one workflow.
Feasibility week · 7 days from receiving your materials$5,000 USD
Credited in full toward the evaluation. We agree how your agent connects on the first call.
- One workflow, with up to 3 of your tool operations mirrored
- Boundary cases on money and authority decisions
- An evidence pack: rule broken, expected vs actual, $ impact, severity, replay
- Rerunnable regression tests and a 30-minute readout
Evaluation · timing agreed in the SOW$20,000 USD
A fuller evaluation across more of your workflow, plus a report written for a vendor-oversight file.
Request an evaluationFor labs · timing agreed in the briefBy quote
A private environment, task set and graders, built to your brief and delivered as a container.
Send a briefWe take on one engagement at a time. We test agents that act through tools or APIs; UI-only agents aren't supported yet.