Matthews and Stream Are Staffing Up to Fix Their Property Data. What Does a Boutique Tenant-Rep Shop Do?

Matthews hired Jimi Li, formerly chief technology officer at ALM, as chief product and technology officer, with product, engineering, data, AI and IT under one executive. Its Artemis system is described as an AI-powered CRM that combines ownership, tenant and comparable-property data with deal tracking and recommendations. Apollo uses property and transaction information to produce valuations, offering memorandums and marketing content. Both draw on a shared company database. The article gives no time savings or adoption figures.

Stream Realty describes Ground Control, a platform of curated property and transaction data, and NavigatorCRE, which combines business intelligence data with maps, dashboards and reports. It is hiring a Director of Market Intelligence & Technology in Florida (a former JLL technology advisory director) and an open Market Research Analyst role in Orange County that maintains property and transaction databases and trains colleagues on research tools. The article says Stream "has not publicly established that generative AI is integrated into every platform" and has not disclosed productivity metrics.

Lee & Associates, in a sponsored article with Bisnow's Studio B, describes its Leo platform. Principal-level brokers check and correct the output, Lee University dedicates its Q4 2026 training to AI, and an eight-person advisory committee led by the CTO reports to executives. Productivity gains are unquantified.

Because the common thread is spending money to keep ownership, tenant, comp and transaction records in one place. At a large firm that is a salary line. At a boutique it is broker hours: the ownership lookup, the prior-tenant check and the comp pull that happen before every tour book.

The Seattle market makes that lookup more valuable, not less. Kidder Mathews' 3Q26 report puts Close-In vacancy at 27.7%, down from 28.2%, with 150,670 square feet of positive net absorption in the quarter but negative 459,019 square feet year to date. Asking rent is $35.02 per square foot, down 6.8% year over year, and sublease space is 5.5 million square feet, 13.4% of available. The report says "both flight-to-quality and flight-to-value trends continue to shape tenant and investor behavior." With tenants holding leverage, the work is choosing the right building, and choosing the right building is a data task.

PLOT Daily reported on October 14 that Solid Earth, an MLS technology provider, says its security software blocks up to 1,200 automated access attempts per day, compared with fewer than ten six months ago. Those are attempts, not confirmed breaches, and the figure comes from one provider.

The other half of the story is the alternative. Flexmls added an MCP server that lets compatible assistants retrieve MLS data through an authorized connection. Flexmls says users cannot retrieve more than their existing MLS access allows, contact and account identity data are excluded, the server does not store or cache MLS data, and each MLS decides whether to turn it on.

MLS is mostly a residential system, so this is a pattern, not a commercial source. The pattern is that data owners are choosing authorized, permissioned connections over open access. Any data layer a small firm builds should assume that.

Three moves, in order of effort.

Keep your own deal history in a structured form: tour notes, landlord behavior, concessions, timelines. It is the one dataset no brokerage platform can buy for you, and it is what makes a tenant rep's answer specific.

Prefer sources that cite their records and state a refresh date, and ask any vendor how it gets its data. If the answer is scraping, ask what happens when the source blocks it.

Scope narrowly. Fisher Brothers, a property owner, caps AI projects at a two-week window "from idea to a deployable tool," per PLOT Daily's October 15 report, and is building reusable MCP connections with permissions, which the article describes as in development. It reports no savings figures. The tenant-rep version is one workflow, such as tour-book assembly, built or bought for two weeks and measured in minutes saved.

None of the three articles reports a tenant-rep result. Apollo is about valuations and offering memorandums, and Lee's article names comps, absorption and ownership records as areas of help for advising investors. Reading these announcements as a threat to boutique tenant reps overreads them.

Hiring is also what firms do before results exist. Matthews, Stream and Lee have published no productivity numbers, so the evidence that these platforms work is thin. And a boutique's edge, an advocate who knows the landlords and the buildings, does not depend on a database at all. The case for a data layer is that it frees time for that edge, not that it replaces it.

List what you re-research for every search: ownership, prior tenants, comps, landlord history. Time each item across your last three searches. That total is the budget a data layer has to beat, and it is the number to hold any vendor to.

That is the gap we are building for at BrokerHQ: a cited building-and-company lookup for firms without a research department. Not every piece is live today, and we will say which ones are when they ship.

• PLOT Daily, "Matthews hires technology chief to expand AI tools for brokers," October 16, 2026.
• PLOT Daily, "Stream Realty builds AI into brokerage research," October 14, 2026.
• PLOT Daily, "Lee & Associates AI platform, broker review and training," October 12, 2026 (sponsored content with Bisnow Studio B).
• PLOT Daily, "AI assistants hit MLS defenses 1,200 times each day," October 14, 2026.
• PLOT Daily, "Fisher Brothers builds AI in two-week projects around legacy systems," October 15, 2026.
• Kidder Mathews, Seattle Office Market Research, 3Q 2026.