Funky.dev builds for developers — and increasingly, for the agents those developers send. The problem was they could not tell the difference in the product.
The problems
Funky was shipping for human developers while a growing share of usage came from agents they could not see.
Invisible agent traffic
A meaningful slice of product usage looked like noise — no attribution, no journey, no way to tell agents from humans.
Broken agent paths
Docs and MCP surfaces that worked for people failed silently for agents, dropping conversion before anyone noticed.
Revenue left on the table
Without knowing who the agent users were, Funky could not recover them or prioritize fixes that would convert.
Why Agentrail worked
Agentrail gave Funky a clear picture of agent discovery, struggle, and task completion — so the team could ship what agents actually needed.
Agent user attribution
Funky finally saw which sessions were agents, where they entered, and where they dropped off.
Journey reconstruction
Failed and successful agent paths became evidence the team could ship against — not guesses.
Docs and MCP that convert
With clear struggle points, Funky improved docs and MCP so agents could complete tasks end to end.
How Funky runs Agentrail day-to-day
Agent journeys stay in the product loop — not a one-off report.
- Watch agent discovery and task completion weekly
- Triage failed agent paths before shipping docs or MCP changes
- Attribute recovered users back to product and GTM decisions
Before Agentrail, we could not attribute agent users, leaving 10% of revenue on the table. Agentrail helped us identify them, recover 15% of lost users, and improved our docs and MCP to drive conversion.
— Jason Jin, Founder, Funky.dev
Results
What changed once Funky could see and fix agent journeys.
15%
users recovered
10%
revenue previously unattributable
Docs + MCP
surfaces improved for agents
