Proptech AISource: NAR
NAR profiles BrokerBot: brokerage-trained AI for policies, CRM admin and workflows
Original headline: “BrokerBot Wants to Rethink How the Modern Brokerage Operates” — NAR, Oct 19, 2026
In an October 19 NAR Broker News interview, BrokerBot.ai CEO and co-founder Jerimiah Taylor — named one of NAR's 2026 Innovators of the Year — described the REACH startup's brokerage-specific AI operating system that connects a firm's policies, forms, workflows and existing tools across web, chat, SMS, voice, email and calendars. Taylor said agents already use consumer chatbots that can expose customer PII or association IP; BrokerBot answers from a brokerage Knowledge Center (uploaded policies, Drive/OneDrive folders, training videos), cites source documents, refuses to guess when knowledge is missing, and says no model trains on the brokerage's information. He positioned payoffs as scaling agent support without adding headcount — from policy Q&A and onboarding to one-click marketing kits and inspection-report workflows that pull approved contractors from the firm's knowledge base.
Our take (opinion & analysis, not reporting)
For broker-owners evaluating ChatGPT sprawl: ask any 'brokerage AI' vendor the same questions NAR's interview raises — does it ground answers in YOUR policy PDFs with citations, refuse when the knowledge base is empty, and keep CRM/PII out of model training? Put BrokerBot on a short demo list next to BoldTrail MCP, eXp MCP and Lofty if you need governed policy answers and CRM admin in one place, but verify folder-level permissions and which live systems it can write to before a firmwide rollout. Agents: if your brokerage adopts something like this, stop pasting listing contracts into consumer chatbots and route those questions through the firm tool instead.
Online discussion: no notable public discussion found yet (checked Oct 20, 2026).