OwnerLM runs a Model Context Protocol server, so the assistant on your desk reads the same brain your team uses: properties, documents, deals, emails, meetings, and the decisions already made. Same permissions as your account, sources on every answer, and every read logged.
$ claude mcp add --transport http ownerlm https://ownerlm.com/api/public/mcpFirst call opens a sign-in and one approval screen. After that the agent works as you. Using Claude in the browser? Settings, Connectors, Add custom connector, then paste the server URL.
Your assistant can do this part itself. Hand it this page, or /api.md if it would rather read plain text, and it can install the server, sign in once, and confirm the connection with a whoami call.
Full setup and tool listAn outside assistant answering from your workspace, with sources attached.
Inside OwnerLM, agents run on the supported model that fits the job. Outside OwnerLM, the assistants your team already pays for read the same approved context. The model can change without moving your records, your permissions, your sources, or your approvals.
How model choice and controls workPull properties, documents, market context, and agent results into the tools you already run: spreadsheets, dashboards, internal apps, your data warehouse, and automation platforms like Zapier, n8n, and Make. Standard REST, JSON responses, keys you issue and revoke in your workspace. Your base address comes with the key, from Settings, Developer.
# $OWNERLM_API_BASE is issued with your key curl -X GET "$OWNERLM_API_BASE/buildings" \ -H "Authorization: Bearer owlm_your_api_key" \ -H "Content-Type: application/json"
const res = await fetch(
`${apiBase}/buildings/${buildingId}/documents?type=mor`,
{
headers: {
Authorization: `Bearer ${apiKey}`,
"Content-Type": "application/json",
},
}
);
const documents = await res.json();An internal tracker your team builds on the same endpoints.
Two ways to build, both reading the same brain and respecting the same rules.
Describe the tool your team needs and it ships inside the workspace, reading the same properties, documents, and agent results.
See StudioPoint your own product at the REST API and skip rebuilding the data layer. Properties, documents, market context, and agent runs are already there.
Read the API referenceThe API and MCP surface does not change with the setup. What changes is where the work happens.
The default. OwnerLM runs your brain and your agents. You connect over the API and MCP from anywhere.
Compute installed at your own property. Same API and MCP endpoints, with work running at or near the property instead of a distant data center, close to cameras, access systems, equipment, and connected property systems.
Routine work runs on your own hardware and heavier work runs hosted. One brain, one set of endpoints, one place your rules and approvals live.
Your data stays yours. Every API and MCP read is scoped to the key you issued and logged, and nothing leaves your workspace unless you choose a data stream.