Last week a user asked if I could put humans on the org chart.
I said sure, and added them. The chart got more useful. Then I put AI agents under the humans, with reporting lines. Even more useful. Then I let humans author SOPs that the agents under them inherit at runtime. That is when the product changed.
OTP started as the Organization Transport Protocol for AI agents. The acronym was right, the noun was right, the scope was wrong. We were treating the agent army as the organization. The agent army is a slice of the organization. The organization is humans and agents, and the SOPs that move between them.
Today the team chart on OTP shows everyone. The CEO at the top, the COO under the CEO, the rest of the human team under the COO, the AI agents wired into whichever human they actually report to. Click any tile, edit anything. Drag any tile, change the hierarchy. Add an SOP to David, watch every agent under David inherit it.
That last part is the load-bearing one. When an AI is spun up under a human who already wrote the SOP for "daily inbox triage," the agent does not have to be re-explained from scratch. It inherits the SOP. The human edits the SOP, the agent updates. The human leaves, the next human picks up where the SOP is, and the agent keeps running. The SOP is the bridge between human work and AI work, and OTP is the substrate it sits on.
This is bigger than what we were doing yesterday. Notion has documents. Lattice has reviews. Process Street has checklists. None of them treat the SOP as an executable spec that an AI inherits at runtime, and none of them have a network where every other organization's SOPs are queryable for comparison.
OTP becomes the layer above all of those. The acronym does not change. The mission expands.
Three things shipped today to start the pivot.
One. Every human and agent on the chart can author SOPs. Title, trigger, steps, outputs, tools, notes. Author once, edit when you learn something, version automatically.
Two. AI agents inherit SOPs from the human they escalate to. There is a "Copy as CLAUDE.md" button on every agent's tile that compiles the agent's own SOPs plus inherited SOPs into a single markdown file you drop into the agent's system prompt.
Three. Five Founder/CEO SOP templates seeded into the editor. Daily inbox, weekly L10, monthly stakeholder update, founder-led discovery call, quarterly Rocks-setting. Click one, copy it in, edit to your voice, save. Sixty seconds to a working operating cadence on the chart.
Next on the build list: invitations so a CEO can pull their team into the chart, claim flow so each member can edit their own tile, and member-scoped permissions. Then a skills taxonomy so the chart can answer "who on this team knows X." Then cross-org SOP comparison once enough orgs are on the network for the comparison to be useful.
The protocol is bigger than I named it. Names follow product. Product follows what users actually do with it.