Marcus Rush, founder and Team Leader of Rush Home, has deployed a custom AI agent named Russ to run portions of his brokerage’s daily workflow, automating lead scoring, agent notifications and calendar tasks for the firm’s eight — agent residential team. Russ operates from its own workspace email and delivers a ranked, actionable morning brief so agents can begin outreach immediately without first logging into the CRM.
At the heart of Russ is a custom lead-scoring algorithm built on Claude that runs across a database of more than 11,000 contacts. The system recalculates scores whenever a new interaction enters its queue, weighing lead source, buyer — versus-seller signals, geographic focus, price points, response timing and call duration. Updated scores are written back into the team’s CRM, Follow Up Boss, and the agent composes a daily brief that ranks leads and pulls suggested follow — up or introduction tactics from CRM notes.
Russ is assembled by layering a custom API integration on top of Follow Up Boss and connecting that layer to Claude for scoring and other services for delivery. Rush used Zapier MCP to create flows that route scoring outputs into Slack for notifications and into Gmail for calendar actions. On building with the platform, he said: "With Zapier MCP, as long as I have the API docs, I can build exactly what I want. We're not limited to a trigger anymore."
Operationally, the automation removes a daily bottleneck: agents no longer must open Follow Up Boss each morning to prioritize outreach. Instead they receive Russ’s ranked list and suggested next steps by email and can start calls and follow — ups immediately. That speed and clarity matter given the volume of leads the brokerage handles and the well-known influence of prompt responses on real estate conversions.
Russ reached full utility once the MCP connections tied scoring outputs into the team’s day-to-day tools; Rush describes the agent as an ongoing project that matured after those integrations went live. He is piloting additional automations: Russ scanning email and managing scheduling and status updates from a dedicated workspace account, which would further offload routine coordination from human agents.
For builders and small teams, Rush’s implementation shows a practical alternative to brittle, trigger — based automation: a model plus API access can enable richer, custom flows. Rush reports that once lead scoring worked he started mapping further use cases — calendar automation, renter profiling and email workflows — illustrating that a custom API layer and a model like Claude can extend automation beyond the limits of traditional no-code triggers.
Sources
Replies (0)
No replies in this topic yet.
