OTP / Bring Your Agents In
Track 2 · You're running agents already · L4–L6

Your agents are running. Now give them a chart.

You've already crossed the starting line. You have agents. They're working. But they're scattered across tabs, scripts, and crontabs, and nobody knows their KPIs.

This track is about formalizing what you've built. Four steps: register the agents you have, give each one a seat on your chart, assign each one a number it owns, and put them on a runtime that doesn't die when your credits run out.

Ollie, the OTP mascot
1

Register your existing agents

~ 20 min

List every agent you're already running. What does it do, who owns it, what triggers it, where does it live today (cron, Claude Code, Make, n8n, Lambda)? This is the inventory step. You can't manage what you can't see.

Open the agent registry →
2

Place every agent on the chart

~ 30 min

Each agent gets a seat alongside your humans. Who does it report to? Who owns it as a manager? If you can't place it on the chart, it's not really part of the org, it's a script. This is where most teams hit the pain wall and realize the chart they had wasn't built to receive agents.

Open the chart →
3

Assign each agent a KPI

~ 25 min

One number per agent. Cold emails sent. Tickets resolved. Loads tracked. Anomalies caught. If an agent doesn't own a number, it doesn't earn its seat, and you can't tell whether it's helping or just consuming tokens. This is the difference between an agent and a script.

Wire up KPIs →
4

Hook up triggers and runtime

~ 30 min

Schedule your agents through OTP's MCP layer. Daily, hourly, on-event. No more cron jobs that die when your credits run out. No more "Dirk stopped working last Tuesday and nobody knew." OTP fires them, watches them, and reports the KPI back to the chart.

Configure the runtime →
When this track gets too small

Track 3 · Coordinate your team

When more than one human is running agents, Track 2's solo-operator shape stops scaling. You'll need multi-user permissions, a cross-agent registry, an inter-agent message bus, and Bassim's L7–L8 maturity scoring. That's Track 3.

See Track 3 →