Pricing

Choose how you work with Agentman.

Have it built for you, build it with us, or build it yourself. Each path is priced for what it is, and each has its own pricing page.

Four paths

One platform, four ways to put it to work.

A clipboard with a medical cross and a stethoscopeBuilt for you

Medman

AI agents for the medical back office: eligibility, prior authorization drafts, inbox triage and phones. Four agents live, three in beta.

from $0.50 / eligibility check
  • Priced by the work each agent does
  • Start with one agent, add the next later
  • Free 14-day trial on every live agent
See Medman pricing

or

A rail of hanging tags, one being hung by handBuild with us

Agentic Harnesses

We build one expert workflow with your team on the platform behind Medman, then operate it with you.

Per engagement
  • Scope, build, operate
  • Your standards as versioned skills
  • Ownership agreed per engagement
A hand taking a book from a row of booksBuild yourself

Agent Skills

Your AI is smart; make it an expert. Write down how your team does the work, and every assistant follows it.

Free to start
  • Free plan: up to 10 private skills and 3 members, no credit card
  • One library for Claude, ChatGPT and Cursor
  • Versions and access control for the team
A hand placing a block on a stack of blocksBuild yourself

Agent Builder

Build, test and run your own agents without code, and see what each one did and what it cost.

Free to start
  • Build and operate agents, no code
  • Session review and cost per agent
  • Team and Enterprise plans as you grow

Agencies, builders and skill authors who deliver on Agentman for their clients have a separate track.

See the Partner Plan
Which one is for me

Start from who you are.

  • You run an independent medical practice

    Medman. The agents already exist. Four are live and three are in beta. You pick the ones you need and pay for the work they do.

  • Your firm has an expert workflow nobody sells software for

    Agentic Harnesses. We build it with your experts on the platform behind Medman, then run it with your team. Scoped and priced per engagement.

  • Your team already uses Claude, ChatGPT or Cursor

    Agent Skills. Write down how your team does the work once, as skills, and every assistant follows it. Free to start.

  • You want to build and run your own agents

    Agent Builder. Build, test and operate agents without code, with session review and cost per agent. Free to start.

Build with us · Agentic Harnesses

A harness is priced as an engagement, not a plan.

There is no price list. We scope one expert workflow with you, build it on the platform behind Medman, and run it with your team. The price follows the scope.

Two magnifying glasses crossed over a document
  1. 1

    Scope

    We pick one expert workflow with you and agree what goes in, what comes out, and what a good result looks like to the people who sign it off.

  2. 2

    Build

    Your team's standards are captured as versioned skills the agents read at run time. Connections, checks and review points are built around them.

  3. 3

    Operate

    The harness runs on the Agentman platform. Your experts review what it produces, and the skills are tightened from what they accept and what they correct.

What an engagement includes

  • Your procedures captured as versioned skills
  • Connections to the systems and documents the workflow needs
  • Evaluation against cases your experts approve
  • Review points and exception routing to the people responsible
  • Operation on the Agentman platform

Ownership, licences and any productization rights are agreed for each engagement.

In production

A debt-restructuring harness is live with a global law firm.

43
specialist agents
96
versioned skills, read at run time
91.3/1001
mean score against partner-authored decks

1 Mean of the four formally benchmarked deals, scored against partner-authored golden-standard decks. Figures as of 17 September 2026, counted from live systems.

Tell us the workflow. We reply by email and set up a scoping conversation.

Built for you · Medman

Medman is priced by the work done.

Four agents are live in independent medical practices and three more are in beta with design partners. Each live agent has its own rate; the beta agents are not priced yet.

Eligibility

$0.50

per check

Prior authorization

$10–$25*

per request

Inbox triage

from $225

per provider / month

Phones

$10.00

per hour of call time

*Depending on complexity. Starting rates; plans, bundles and the free 14-day trial are on the Medman pricing page. Medman is built for HIPAA, and we sign a Business Associate Agreement before any patient data moves.

See all Medman pricing
Questions

Not sure which path fits?

Which option should I start with?

A medical practice starts with Medman: the agents are already built. A firm with an expert workflow that no product covers talks to us about an agentic harness. A team that wants its own assistants and agents to follow its standards starts free with Agent Skills or Agent Builder.

How much does an agentic harness cost?

There is no price list. A harness is a scoped engagement: we agree one workflow, build it with your experts, and operate it on the Agentman platform. The price follows that scope. Ownership, licences and any productization rights are agreed for each engagement.

How is Medman priced?

By the work each agent does, not by seat. Eligibility starts at $0.50 a check, and each of the four live agents is priced on its own, so you can start with one. The three agents in beta are not priced yet. Every live agent has a free 14-day trial. The Medman pricing page has every rate and plan.

What does the free Agent Skills plan include?

Up to 10 private skills and 3 members, no credit card. Paid plans add more room and more control as a team grows; the Agent Skills pricing page lists them.

What is the difference between a skill, an agent and a harness?

A skill is a written procedure: how your team does one piece of work. An agent follows skills to do that work. A harness is the operating environment around one or more agents: the tools, context, saved progress and checks that let them carry a whole workflow, with people reviewing where it matters.

Do skills work across Claude, ChatGPT and Cursor?

Yes. Your library is served over one connection, so the same skills are available in each assistant a teammate connects.

Next step

Tell us the work. We will tell you the path.

If none of the four is an obvious fit, describe the workflow and we will point you to the right one.