Open OS Meets Open Agent: Leveraging FUB's Integrations Inside a Single Ace AI Prompt

Why does Follow Up Boss have so many integrations?

Follow Up Boss was designed from the start as an open platform rather than a closed system. The philosophy is that a real estate team already uses a specific dialer, a particular IDX website provider, and preferred lead portals — FUB should fit into that stack, not replace it. The result is 200+ documented integrations spanning lead portals, email marketing, Google Workspace, dialers, transaction management, and more (per FUB's own published integration marketplace at followupboss.com/integrations).

The practical consequence: every lead source that pushes into FUB — Zillow, Realtor.com, RealScout, your IDX site — can be read and acted on through a single system of record. That matters enormously when you layer an AI agent on top.

What does "open agent" mean for a real estate CRM?

An open agent is an AI model — Claude, ChatGPT, or any MCP-compatible assistant — that can take actions inside external systems rather than just answering questions. For FUB users, that means the AI can read contact records, check pipeline stages, flag compliance issues, and surface coaching cues directly inside the CRM through structured tool calls, not copy-paste workflows.

Follow Up Ace exposes this capability through the Model Context Protocol (MCP): a standardized way for AI assistants to call external tools over a single authenticated URL. Two connector endpoints are active today (verified in chat-app/routes/embed.js):

• Claude (Streamable HTTP): https://followupace.com/mcp
• ChatGPT (full streamable HTTP): https://followupace.com/mcp?tools=full

Paste either URL into the respective AI client's connector settings and the assistant gains access to 189 MCP tools across your FUB account — the full connector surface documented in the MCP server.

What can a single Ace AI prompt actually do across FUB integrations?

Once the MCP connector is active, a single natural-language prompt can chain together multiple tool calls that previously required four or five manual navigation steps. The 189 MCP tools span contact management, pipeline analysis, lead-source performance, compliance scanning, and Zillow follow-up workflows. Here are four concrete examples drawn from tools verified in the codebase:

1. Pipeline health check

The pipeline-health-check tool (MCP server index.ts line 4198) runs a comprehensive analysis of your sales pipeline — stale deals, stage velocity, conversion gaps — and returns a structured report. One prompt, no manual filtering through FUB smart lists.

2. Speed-to-lead audit

The lead_speed_to_lead_check tool (verified in mcp-server/src/zillowTools.ts) checks whether a specific contact is still inside the critical 5-minute response window and returns average response time, percentage within the target, and trend data. Ask "Is Maria Chen still in my speed-to-lead window?" and the agent answers with numbers, not a guess.

See the Zillow Playbook for the full speed-to-lead workflow this tool supports.

3. Lead nurture optimizer

The lead-nurture-optimizer tool (MCP server index.ts line 4230) identifies which contacts in your pipeline are at risk of going dark and surfaces the recommended next action per contact. It works across every lead source that feeds FUB — the integration source does not matter because the data is already unified inside the CRM.

4. Fair Housing compliance scan

Before any AI-generated message leaves the platform, the scanForComplianceViolations() function (verified in chat-app/utils/complianceGuard.js line 293) checks the text against Fair Housing rules and licensing requirements. The agent will not send or surface language that triggers a violation. See the Compliance page for what the scan covers.

How does the Ace Win Score work inside FUB custom fields?

Even without the MCP connector, completed supported scores can be delivered to native FUB custom fields. Webhook receipt can schedule work, but it is not proof that analysis or write-back completed. The field system uses two tiers:

Once the definitions are verified and values are present, agents can sort or filter them inside FUB. Live explanations and supporting evidence remain in Companion and the dashboards. Learn more about Ace Trove.

How do you connect Ace's AI agent to Claude or ChatGPT?

Connecting takes under two minutes. The steps below assume you are already on an active Follow Up Ace account and have your MCP URL available in your account settings.

1. Open Claude Desktop (or the Claude web interface with MCP support) and navigate to Settings > Integrations.
2. Add a new integration and paste the Streamable HTTP connector URL: https://followupace.com/mcp.
3. Authenticate with your Follow Up Ace credentials when prompted.
4. In Claude, type a prompt referencing your FUB contacts or pipeline — for example, "Run a pipeline health check and tell me which deals haven't had contact in 14 days."
5. Claude will call the relevant Ace MCP tools, retrieve live data from FUB, and return a structured answer in the chat window.

For ChatGPT, use the streamable-HTTP URL (https://followupace.com/mcp?tools=full) in ChatGPT's plugin or action connector settings. The workflow is otherwise identical.

What FUB Smart List recipes does Ace unlock?

Beyond one-off AI prompts, teams can turn the verified fields into repeatable FUB review views. The model narrows attention; an agent or admin still reviews the underlying record before taking action.

• Priority review — sort numeric Ace Win Score descending; it is a within-account rank, not a temperature
• Churn save-list — filter Ace Churn Risk = High, then inspect native recency and communication history
• Active buyer review (Trove) — filter Ace Opportunity = Active Buyer and optionally Buyer Tier = Very High or High
• Seller review (Trove) — filter Seller Tier = Very High or High and inspect Seller Radar evidence before outreach

Last Analyzed provides a freshness date for Trove output. A missing value remains unknown; do not coerce it to zero or a low tier.

What does Ace cost compared to managing integrations manually?

Ace operates on three per-seat plan tiers (verified in chat-app/routes/embed.js and chat-app/routes/zillow.js):

Teams that need account-wide AI analysis — nightly Seller Radar scans, warehouse-backed revenue analysis, bulk contact scoring across thousands of contacts — add the Ace Trove, which is priced by contact count from $49/mo (Starter, up to 5,000 contacts) to $899/mo (Enterprise, up to 500,000 contacts).

The manual alternative — stitching together Zapier automations, a separate analytics dashboard, and a standalone AI tool — typically costs more in both subscription fees and agent time spent switching between systems. The MCP connector eliminates the switching by making FUB queryable through whichever AI assistant the agent already uses.

What makes FUB's open architecture different from a closed CRM?

Closed CRMs lock data inside proprietary formats. Every integration requires the vendor's blessing, and AI features are limited to whatever the vendor ships. FUB's open architecture inverts that relationship: because FUB exposes a public API and accepts leads from any source that sends an email or posts to a webhook, the data is already interoperable before any AI layer arrives.

Ace takes that openness one step further by exposing the same API surface through the MCP standard. That means a team can use Claude today, switch to a different AI model next year, and the connector URL stays the same. The AI model is swappable; the CRM data and the tool definitions are not.

This is what "open OS meets open agent" means in practice: FUB is the open operating system for real estate data, and the MCP connector is the open interface that lets any compliant AI agent run inside it.

Which FUB integrations benefit most from an AI layer?

Not all integrations benefit equally. The highest-leverage combinations are those where the AI can act on data that already exists in FUB because an integration pushed it there:

• Zillow / Realtor.com lead feeds — AI checks speed-to-lead window the moment a lead lands; no manual triage needed.
• IDX websites (RealScout, Showcase IDX, etc.) — search behavior can enrich live property context in Companion without creating a stale Search Area FUB field.
• Gmail / Google Workspace — email activity tracked in FUB can become model evidence where supported, while the original communication remains available for review.
• Native FUB calls and texts — channel and response history provide context for a Win Score or Churn Risk review; they are not additional Ace fields.
• Transaction management (Dotloop, SkySlope) — when a deal moves to a transaction stage, the AI can shift the contact's Follow Up Boss Stage to Active Client and adjust follow-up cadence automatically.

In each case, the AI is reading data that FUB's integrations already collected — it is not duplicating work, it is making the existing data useful faster.

Does the AI agent work with Zillow leads specifically?

Yes. The Zillow speed-to-lead check is one of the most frequently used tools in the MCP surface. When a Zillow lead lands in FUB, the lead_speed_to_lead_check tool can be called immediately to determine whether the agent is still within the 5-minute contact window that Zillow's own research identifies as the threshold for meaningful connection rates.

The tool returns three data points: average first-response time, the percentage of leads contacted within the 5-minute target, and a trend indicator. If the agent is trending slower, the AI can escalate to a pipeline health alert without any manual report generation. See the dedicated Zillow Playbook for the full recommended workflow.

How do I get started with the MCP connector today?

1. Create a free Follow Up Ace account at followupace.com/admin/signup — no credit card required.
2. Connect Follow Up Boss in one click from the onboarding screen. Ace reads your FUB account's contacts and pipeline immediately.
3. Copy your MCP URL from the Ace settings panel (Claude URL or ChatGPT streamable-HTTP URL depending on your preferred AI assistant).
4. Paste the URL into your AI client's connector or integration settings and authenticate.
5. Verify field readiness. Have the FUB account owner authorize custom-field creation, then confirm structural verification before using Win Score or Churn Risk in a live view.

For more context on how the agentic layer fits into a full real estate workflow, see the Agentic overview and the Follow Up Boss integrations guide.