Lofty, the Phoenix-based software company formerly known as Chime Technologies, said Sept. 28 that it has released an MCP server for its real estate platform — a piece of plumbing that lets a brokerage point its own artificial intelligence tools at the data and workflows it already keeps in Lofty.
The announcement matters less for what it adds to Lofty’s own product than for what it concedes about how brokerages are buying AI. Rather than asking firms to run everything through one vendor’s assistant, Lofty is opening the platform to whatever AI clients a brokerage chooses.
“Lofty brings together a brokerage’s people, data and workflows into one operating system, and now we’re making that intelligence securely accessible to the AI agents our customers choose to use,” Chief Executive Officer Joe Chen said in the announcement.
What an MCP server actually is
The Model Context Protocol is, in the project’s own description, “an open-source standard for connecting AI applications to external systems” — databases, files, search tools and defined workflows. The protocol’s documentation compares it to a USB-C port: a single standardized connector, so a tool built once can plug into many applications. A server exposes a system’s data and capabilities; a client is the AI application that connects to it.
In practice, that means a brokerage running its own AI assistant, or using a general-purpose one, can have that assistant read information out of Lofty, analyze activity across it and — where the brokerage has authorized it — take actions inside the platform, without a developer writing custom integration code for each connection. Lofty said the server keeps the controls and authentication required for business use.
Lofty did not disclose pricing, a general-availability date, or which brokerages are using the server.
A year of building the stack
The release is the third public step in a product line Lofty started in the winter. In February the company introduced Lofty AOS, an agentic operating system built around a set of named modules including an AI Assistant, Sales Agent, Social Agent, Homeowner Agent, Website Builder, SEO/AEO Manager and Transaction Coordinator. Chen said at the time that “the future of real estate belongs to organizations that move beyond simple AI tools and embrace agentic systems.”
In July the company expanded AOS around a workspace called Lofty Cowork, with features named “Your Morning Read” and “What Needs You Now,” plus customizable agents and copilots. Chief Technology Officer Henry Li framed that release as an answer to the adoption problem that has dogged real estate software: Cowork, he said, “eliminates the dependence on agent adoption to make AI a reality.”
The MCP server points in the opposite direction from a closed suite. It is an admission that the larger brokerages — the ones with engineering teams, proprietary data and their own model deployments — will not hand the whole job to a single vendor, and that a platform’s value increasingly lies in being reachable by other systems.
Where brokerages are on AI
Vendor activity has run well ahead of daily practice, though the gap is closing. The National Association of Realtors’ 2026 technology report found that nearly half of Realtors now use AI at least weekly.
Competing products have taken different routes to the same customers. Lone Wolf Technologies built an AI assistant into its BrokerMetrics analytics platform to flag recruiting prospects and agents at risk of leaving, and FirstTeam Real Estate adopted an outside AI operating platform from Purlin rather than build its own. What distinguishes the Lofty announcement is that it does not try to be the destination: it is an interface for brokerages that have already decided to run AI on their own terms.
More technology and AI coverage is available from RealtyWire.
