You probably already have an AI agent that knows how you write, what you care about, and what you are trying to build. It lives in the app you open every morning. It is good at conversation and good at thinking things through with you.

Marketing is a different job. It needs your brand context held in one place, specialists for each channel, a steady cadence, and a clear point where a person says yes before anything goes out. A chat window is not built to carry that for months.

So we did not build another assistant to compete with yours. We made Lane something your agent can work with.

Your agent is the remote control

Lane is an AI brand automation platform. Lane runs the marketing loop in its own cloud: it reads your brand foundation, finds what is worth testing, drafts the work, and lines it up for review.

When you connect your own AI agent, it gets a way into that system. You sign in to authorize it. After that, your agent can list your brands, read a brand’s setup (foundation, objectives, personas, and voice), and make edits to a brand’s setup and to drafts waiting for you. Those edits apply immediately and replace what was there, so it is worth checking what your agent changed.

Think of it as a remote control. The work runs in Lane. Your agent lets you check in and steer from the place you already talk to.

What stays with you

This is the part we care about most. A connected agent cannot:

approve anything
publish or send anything
spend money
delete anything
change your budgets or Lane’s run settings
choose who an outreach email goes to

Approval tools are not even visible to a connected agent. You approve in the Lane Actions queue in the app, and recipients are chosen only by you, in that queue. Nothing is published, sent or spent until you approve it there.

That boundary is deliberate. The more capable agents get, the more it matters that one person can see what is about to happen and decide.

What your agent sees

Whatever your agent reads from your brands becomes part of your conversation with it, so your agent provider’s own terms and privacy policy apply to that data. For Lane, we do not sell your data, and we share only the data necessary with service providers and connected integrations to operate the agreed services. Details are in our Privacy Policy.

You can end access at any time by removing the Lane connector in your agent or signing out of it.

What we have tested

Lane offers one remote MCP (Model Context Protocol) server, and an agent connects to it with sign-in. We name an agent only after testing it end to end. So far that is Claude, ChatGPT and Hermes. Claude and ChatGPT each connect on your own: add https://mcp.luminarylane.app/mcp as a custom connector, sign in, and approve the connection on one page. Hermes connects to the same address with its own sign-in. We will add others to the list only after we have tested them.

Where to start

If you already use an agent day to day, the quickest way in is the step-by-step guide.

If you are new to Lane, read how Lane works, see what your agent can and cannot do, or get started.

Your agent already knows you. Lane gives it a brand team.