OperationsCloud brokerage: the client journey and the stack behind it
Seven stages from stranger to repeat client, every AI and human touchpoint along the way, and a named, buildable tech stack for a brokerage that stays lean as it scales.
About
A cloud brokerage has no offices to hide behind, so the client experience is entirely made of software and a small number of people used carefully. This maps both halves at once.
The top row is the journey: discovery, capture, the first five minutes, the long middle where most leads actually live, the buyer and seller paths (now legally distinct in a way they were not before 2024), the run to closing, and the after-close period that produces the cheapest leads a brokerage will ever get.
The middle row is the stack, with real product names in it. A website on Next.js with Sanity for content and a RESO Web API feed for listings. Follow Up Boss as the CRM, with the reasons you might refuse it. A thin AI layer behind a model gateway. SkySlope or Brokermint for transactions. Twilio for messaging, with the registration work that implies. Postgres as the system of record, because the CRM must not be.
The bottom row is the part that decides whether it works: the four human roles worth paying for, the legal guardrails that cannot be designed around, and five ways this shape fails in practice.
The organising idea is that AI belongs at the two ends and in the drafting, never in the middle. It should answer in seconds at 11pm, keep a dormant database warm for eighteen months, and prepare work so a licensed human can do the judgement quickly. Everything involving price, advice, negotiation or a signature stays with a person.