AI skills platform Agentman says Medman cuts eligibility-check costs 94%, discloses no insurance terms of its own

Agentman, which sells reusable AI "skills" for use inside Claude and ChatGPT, says its Medman product is running in medical practices today, automating insurance-eligibility checks and claims workflows. The company describes Medman on its site as "a production harness coordinating eligibility, inbox triage, and claims loops across 1,300+ payers, with the audit trail to show what every agent did."

Agentman publishes specific performance figures tied to the deployment: "90% eligibility automation in 12 weeks," "65% fewer eligibility denials," and a cost of "$0.50 per eligibility check vs. $7.97 CAQH benchmark" — a reference to the manual-verification cost estimate published by the Council for Affordable Quality Healthcare. The company says results so far come from "2 agents live, 6 on the 2026 roadmap."

Sachin Gangupantula, co-founder and vice president of practice operations at Valley Diabetes & Obesity, is quoted on Agentman's site describing the effect on staff workload: "We cut 50 hours of monthly verification calls down to 7. Staff aren't drowning in eligibility calls anymore — they're actually talking to patients."

The company's broader platform is built around what it calls "skills" — versioned, permissioned procedures meant to capture how a team's most experienced staff already do the work, reusable across different AI tools.

Under AI Dispatch's five-dimension trust-gap rubric, Agentman's published material supports a real audit-trail finding: the company specifically ties its claim to Medman's production harness logging what it did. Insurance/indemnity, outcome-settlement and liability-cap terms are absent from the same site — fields AI Dispatch's index records as no public information for the company. Medman operates in a regulated, high-risk domain, medical eligibility and claims, where an automated error creates liability for the clinic using it, not necessarily for Agentman. The cost and denial-rate figures the company cites come from its own case study measured against a third-party benchmark; AI Dispatch could not independently verify them.