AI agents for commercial real estate firms

Commercial real estate websites are built around listings, and listings generate the same questions over and over: is it still available, what is the rate, what are the expenses, how much parking, can I see it this week. The answers change month to month, and the people asking range from a startup founder looking for 3,000 square feet to a tenant rep with a client's full requirement. This page covers where AI agents help a commercial brokerage, landlord or property manager, and the limits that matter in this business.

Who visits a commercial real estate website

• Tenants and their brokers. Space by size, submarket, use and budget. They want availability, asking rate and lease type, operating expenses, parking, power, ceiling or clear height, loading, and a tour.
• Investors. Properties for sale, offering memoranda, rent rolls and operating history, usually after signing a confidentiality agreement.
• Property owners. Leasing and management services, what you charge, and what you have done for similar buildings.
• Cooperating brokers. Availability, commission terms and tour logistics.
• Current tenants of buildings you manage, looking for maintenance requests or building rules.

Which ASDK agents fit

The agents share one knowledge base built from your website: listing pages, building pages, service pages and policies.

Sales: qualify requirements and book tours (early access)

Sales turns "do you have anything like this?" into a requirement your broker can act on. You set the questions in plain language: size range, location, use, budget, move-in date, current lease expiration, and whether the visitor already has a broker. Qualified requirements get a tour or call booked through Cal.com, Calendly or Google Calendar, and the lead goes to your team by email or Slack, and to your CRM through webhooks and Zapier or Make.

Example flow

1. A visitor writes: "We need 8,000 to 10,000 square feet of office near the airport, ideally by spring."
2. Sales asks for headcount, budget per square foot, required move-in date, when their current lease ends, and whether a tenant rep represents them.
3. Two of your listings match on size and submarket. Sales links both, notes the "as of" date on each, and offers tour times with the listing broker.
4. The visitor says they have a broker, so Sales captures the broker's contact details and routes the requirement to the cooperating-broker workflow instead of pitching directly.

Chat: answer listing and building questions (early access)

Chat answers from your listing and building pages and links the source, so a visitor gets "yes, and here is the flyer" at 9 p.m.

Example flow

1. A visitor asks: "Is suite 210 at 400 Main still available, and what's the rent?"
2. Chat answers from the listing page: available as of the date shown there, the published asking rate and lease type, and a link to the flyer.
3. It adds that availability and terms are confirmed by the listing broker and offers to book a tour.

Agent-Ready: make listings readable to AI agents (early access)

Agent-Ready publishes an llms.txt file that lists your markets, property types and current listing pages, so an AI system lands on the live listing rather than an old flyer. A tenant rep's agent can "ask a question" about a suite's size or lease type, or use "contact sales" to request a tour with the listing broker. On Pro and above, visit analytics show which AI agents read which listings.

Guardrails for commercial real estate

A chatbot's answer is your firm's answer. In 2024 a British Columbia tribunal held Air Canada responsible for what its website chatbot told a customer, noting that it makes no difference whether information comes from a static page or a chatbot (Moffatt v. Air Canada). In commercial real estate, set these limits:

• Availability and pricing only from current listings, with the date shown. Remove leased or withdrawn space from the site, and the agent stops mentioning it.
• No negotiation. The agent does not agree to rates, free rent, tenant improvement allowances or lease terms. It collects the request and passes it to your brokers.
• Ask about representation early. Whether the visitor already has a broker changes who should talk to them next.
• Confidential documents stay behind your process. Offering memoranda, rent rolls and operating statements go out after a confidentiality agreement, never through chat.
• Keep private offerings off the public agent. If your firm sponsors investment offerings under Rule 506(b) of Regulation D, that exemption does not allow "general solicitation or advertising to market the securities" (SEC). Keep those offerings out of the public agent's knowledge base entirely, and have securities counsel review anything investor-facing.
• No projections or investment advice. The agent does not estimate returns, cap rates for a buyer's plan or future rents.
• Square footage exactly as listed, and say which measure it is. Rentable and usable area differ, and BOMA's office standard exists mainly to calculate rentable area (BOMA), so a rate quoted per square foot means little without the basis.

How AI agents research space and service providers

A corporate real estate manager or a tenant rep can ask an assistant: "find 20,000 square foot warehouses in Phoenix with 32-foot clear height and at least four dock doors, available within six months." Assistants such as ChatGPT, Gemini and Claude search the web and read listing pages. Agentic browsers can go further and fill in an inquiry form or request a tour. A listing that exists only as a PDF flyer or an image is harder for them to read and compare than a plain page with the facts written out.

Publish these as text on each listing page:

• Size and divisibility, with the measurement basis.
• Asking rate and lease type (triple net, modified gross, full service), or "contact for pricing" if you do not publish it.
• Availability date and the date the listing was last updated.
• Building specifications: clear height, loading, power, parking ratio, zoning, year built or renovated.
• Submarket and address, with property type and permitted uses.
• Who to contact and how to book a tour.

At the firm level, publish your markets, property types, services for owners and how cooperating brokers work with you. Listing platforms and map widgets often load listings with scripts, which leaves an AI agent with an empty page; the free agent readiness checker shows whether your listing text is in the HTML. The agent-ready website checklist covers structured data for your offices and llms.txt, and Agent-Ready keeps the summary current as space leases up.

Where to start

Start with Chat on your listing pages and Sales on your "find space" and contact pages, with the guardrails above in their instructions. Review the first week of conversations: questions the agent could not answer usually mean a listing is missing a fact tenants care about. Pricing shows which plan includes each agent.