All field notes

What is PreMan? Infrastructure for the Agentic Era

PreMan helps keep software reliable as AI agents become its users, with parallel MCP probes, sandboxed workflow scenarios, and repeatable tests.

On this page6 sections
  1. 01Software has a new kind of user
  2. 02Simulate relevant conditions
  3. 03Probe MCP connections in parallel
  4. 04Turn failures into repeatable tests
  5. 05Who PreMan is for
  6. 06Where we are going

PreMan is infrastructure to keep software reliable as AI agents become its users. It helps teams simulate agent behavior, probe MCP tools in parallel, and turn workflow failures into tests they can run again.

We build for teams shipping software used by agents alongside people. As those users change, the workflows your team needs to test change with them.

Software has a new kind of user

Agents choose tools, combine actions, and take paths your team may never have written a test for. A successful tool call can still lead to the wrong business outcome.

For example, a refund might succeed even though the agent skipped the return required by company policy. To check that workflow, a team needs to examine the decisions between calls and the state left behind.

Simulate relevant conditions

PreMan recreates the conditions that matter in a sandbox. Teams bring their tools, workflows, and business rules, then describe the expected behavior in plain English.

That gives a scenario a clear goal and expected outcome: what an agent should accomplish, which actions it can take, and the rules it must follow.

Probe MCP connections in parallel

An MCP server needs to handle agents connecting and discovering its tools. PreMan can send concurrent simulated agent traffic and make the request and response evidence available for inspection.

The homepage demo shows the Harness matrix swarm setup: choose agent environments, set the swarm size, and control how many agents run at once. Workflow scenarios add checks on the decisions agents make while using software.

Turn failures into repeatable tests

When a workflow fails, keep the scenario, expected outcome, and evidence. Run that check again after a tool, model, or product change.

PreMan also helps teams discover existing tests, map the workflows they cover, and add adjacent scenarios. Developers can run and expand tests from the desktop app or terminal using plain English.

Who PreMan is for

PreMan is built for software teams that ship quickly and need to understand how agents use their products. That includes teams building MCP tools, maintaining agent workflows, or expanding coverage as their products change.

Start with a workflow your business depends on. Define its rules, exercise the paths agents can take, and retain useful checks for the next release.

Where we are going

We believe fleets of agents will become major users of software. Our ambition is to build the infrastructure that keeps products reliable in a world where millions of agents use them alongside people.

Today, PreMan tests software and agent workflows, including parallel MCP probing. Broader fleet workflows are part of what we are building toward.

→ Book a demo

Build for agent users

Your product shouldn’t break when agents use it.

Simulate agents, probe MCP tools in parallel, and turn workflow failures into repeatable tests.