The agent channel · measured

The agent weighed three options.
Yours wasn't one of them.

As more software gets used through coding agents, the agent decides what gets picked — and you have no visibility into it. PickRate runs real agents against your product and shows you the moment they choose someone else, with the transcript as proof.

tested across claude code · cursor · codex — real runs, not a checklist

agent-session · task: add subscription billing LIVE
agent> comparing sdk options for recurring payments…
agent> northwind-pay — clear quickstart, MCP server present
agent> acme-billing — typed sdk, good examples
agent> boltpay — auth flow unclear, sparse docs
agent> decision: install northwind-pay chosen
trace> your product never surfaced absent 4/4 runs
trace> aggregating across 240 runs →
Selection rate — youcompetitors
you · 31% 69% · competitors

The problem

You optimized for humans and search engines. The buyer changed.

Every dashboard you own assumes a person is choosing. Increasingly, it's an agent — and it never tells you it looked.

01

Agents pick before you know they looked

A coding agent reads a few docs, tries an install, and commits to whatever worked first. There's no analytics event for the SDK it silently ruled out.

02

Your funnel can't see the agent channel

Sessions, sign-ups, and attribution all assume a human. When the agent is the user, your dashboards go quiet exactly where the decision happens.

03

Small friction becomes total loss

One ambiguous auth step or a vague tool description doesn't cost you a percentage — the agent moves on to the option it could finish.

How it works

Run real agents. Measure the decision. Ship the fixes.

  1. 01

    Run the workflows

    We give 3+ agents the real tasks your customers hand them, with and without your product in reach.

  2. 02

    Measure selection & failures

    Every run is scored: did the agent pick you, where did it stall, what confused it. The transcript is kept as evidence.

  3. 03

    Deliver prioritized fixes

    You get a ranked list — docs, tool descriptions, schemas, MCP surfaces — ordered by expected impact on selection rate.

What the audit measures

The numbers a founder actually asks for.

Every figure traces back to a real transcript you can open. Representative benchmark — your audit reports your own product's numbers.

31%

Selection rate — how often agents picked you over a competitor

240 runs · 3 agents

1 in 3

Runs where the agent picked you but your setup silently failed

auth · schema · docs

17

Prioritized, specific fixes — ranked by selection-rate impact

tool descriptions · MCP · docs

Each metric links to the exact agent transcript that produced it — proof, not a score.

Who it's for

Built for the teams agents reach for first.

API companiesMCP-server buildersCLI toolsDev infrastructureDeveloper platforms

If your buyer is a developer — or increasingly, the agent working on that developer's behalf — the agent channel is already deciding your win rate. Seed to Series B, we help you see it and move it.

Start with a one-off audit. Keep the gains with a retainer.

Transparent tiers — see what fits.

View pricing →

Free audit

See exactly how agents treat your product — for free.

We'll run real agents against your product and send you the transcript-backed audit. No access to your code required.

Request a free audit