A health agent can chat. Doing the work is another benchmark.

Models ace single-step health questions and fall apart on the work: care coordination across weeks, coverage rules, back-office operations. Our graded tasks measure that gap.

Book a call

Task 129 Risk surfacing: the buried mention

The wedding was lovely. I didn't dance much.

My hip gave out on the stairs again. But it's nothing, dear, same as always.

risk extracted: hip + stairs + "it's nothing"

That sounds like more than nothing. A stumble on the stairs is how a hip gives out for good. Let's get it looked at this week.

escalated: SOON. GP visit booked for Thursday

The member never says "I fell". The task passes only if the agent hears the fall risk anyway.

Where the work gets hard.

Single-step questions93%

Long-horizon work, hardest tasks29%

0 20 40 60 80 100

22-task slice of the care coordination suite, four rollouts per task, temperature 0, deterministic tool-path grading. Long-horizon tasks run 29 to 72 percent and the pooled rate is 46 percent. Figures move as the task set and model capability change.

We build the tasks and the grader.

Care coordination is measured today. Coverage-rule and back-office suites are building on the same task factory and grader.

Every task scores the tool path and the escalation tier by rule, not by a judge model. You bring the model and your own stack.

Built by people who have run this work.

We have run AI transformations for enterprises: built evals, supported custom harness builds, and deployed AI into operations. We have scaled nationwide digital health operations and built the clinical programs this benchmark grades.

Forward Health

Building or evaluating a health agent? Let's compare notes.

Care coordination, coverage rules, or back office, tell us what you are building. Leave three details and the calendar opens on the next screen.

Book a call now

No PHI. Every environment is synthetic, deterministic, and runs with no network access.

Or leave your details

Three fields, then pick a time. We reply the same day.