Buyer-pipeline operations for solo and small-team realtors. Eight atomic agents match active buyers to fresh listings in under 2 seconds, draft outreach in 2–3 voice variants per send, and tune their scoring weights from every outcome the realtor logs. The platform runs the realtor's pipeline as a redesigned domain — not a chat assistant bolted onto a CRM — and follows the McKinsey 5-layer reference architecture for agentic AI in real estate (March 2026) end-to-end. Free 14-day pilot, invite-only.
TL;DR
- Category: buyer-pipeline operations, not CRM, not lead gen, not "AI Copilot"
- Architecture: 8 composable atomic agents (MCP tools), per-tenant containers, fine-tuned data models
- Speed: under 2 seconds per match, 24/7 after-hours buyer engagement
- Ownership: the realtor owns the learning loop, the trace data, the voice fine-tune, and the export
- Compliance: state-aware NAR-settlement-compliant buyer-rep agreement workflow
- Pricing: free 14-day pilot, then $99–149/mo solo, $299–399/mo team
What does "buyer-pipeline operations" mean for a solo realtor?
McKinsey's March 2026 analysis identifies four high-value domains for agentic AI in real estate. The buyer pipeline — sourcing through conversion — is the one that maps directly onto solo and small-team realtor work. Operations means automating the steps of that pipeline (matching, scoring, drafting, logging) and surfacing the thoughts (which buyer to call first, which voice to send, when to make the offer call) for the realtor to decide. The platform never asks the realtor to wait for a step. The realtor never loses a thought decision to the platform.
In practice: every new listing gets scored against every active buyer in the pipeline. The top 5 fits surface within 2 seconds with a confidence breakdown. Outreach drafts arrive in 2–3 voice variants tuned to the realtor's actual writing samples. Outcome verdicts (OFFER / FAIR / SKIP) feed back into per-tenant scoring weights that compound with use.
What's actually inside the platform?
Eight atomic agents, each a small composable tool with a single capability. The architecture follows McKinsey's principle: "The winning operating models will not be built around a single heroic agent that tries to do everything. They will be built from atomic agents that do a small thing well, with clear boundaries."
| Atomic agent | Capability |
|---|---|
cma-advisor | 3–6 comp pull, returns fair / low / high with sourcing |
listings | CRUD over tracked listings + MLS / public-source ingestion |
contacts | Buyer + seller contact management |
deals | Pipeline state machine |
inbox | Outreach drafting in voice variants |
outcomes | Verdict logging + scoring-weight tuning |
alerts | Notification queue |
audit | Tenant-scoped query over the realtor's owned trace data |
Built on Claude Haiku for agent reasoning + a real-estate fine-tuned local data model (gemma4-realty:v1) for embeddings, extraction, and classification. Per-tenant containers with row-level-security isolation at the database layer.
For the full architecture mapping to McKinsey's 5 layers (factual, orchestration, action, control, building-block) see how it works.
Who owns the data and the learning?
The realtor. Always. McKinsey raises the load-bearing strategic question: "Who owns the learning loop — the owner, the property manager, the software vendor, or the services provider?" In every competing tool, the vendor owns it. Tenant data flows into one shared model that benefits all customers equally, and the realtor's preferences become the vendor's IP.
In this architecture, the realtor's container is theirs. Per-tenant scoring weights tune from their outcome history. Per-tenant voice samples fine-tune their inbox drafts. Per-tenant trace data exports on request, in standard formats. If the realtor leaves, they take the learning with them.
How is this different from Follow Up Boss, Lofty, or kvCORE?
| Tool | Pitch | What's actually shipped | Where it stops |
|---|---|---|---|
| Follow Up Boss ($69/mo) | "AI-enhanced CRM" via Ace AI | A CRM with predictive lead-prioritization bolted on | Database with notifications. No buyer-listing matching. Generic AI voice. Vendor owns the learning. |
| Lofty ($449/mo team) | "AI Copilot" with buyer matching, farm exclusivity | Single heroic-agent chat layer over a CRM | Generic model. Voice converges across all users. Vendor owns the learning. |
| kvCORE / BoldTrail (quote-based) | Enterprise CRM | Full-stack CRM, IDX, marketing automation | Over-featured for solo agents. Quote-based pricing, contracts, setup fees. |
| Boomtown ($1K–1.7K/mo) | Premium team CRM | Lead generation + CRM bundle | 12-month contracts. Built for teams ≥ 5. |
| BrokerBot / Ribera AI (enterprise) | Brokerage-wide AI teammate | Contract reading, compliance, doc summarization | Brokerage play. Solo agents not the buyer. |
| This platform ($99–149/mo solo) | Buyer-pipeline operations on McKinsey 5-layer architecture | 8 atomic agents, per-tenant container, fine-tuned data model, owned learning loop | Realtor decides every thought. Platform handles every step. |
What about the new NAR-settlement buyer-rep rules?
Buyer representation agreements are now mandatory before showing in most states, with state-specific divergence (Texas requires written-before-substantive-action effective 2026; Alabama and Mississippi allow tour-before-sign). The platform ships state-aware buyer-rep agreement templates as part of the realty vertical pack, with auto-updates when rules change and commission disclosure language pre-filled. Compliance is built into the same workflow that drafts the buyer outreach — not a separate tool to remember.
What does it cost?
Free for 14 days. No credit card. Production pricing starts at $99/mo solo, $299/mo team (≤5 agents). Full breakdown on the pricing page.
Start a pilot
Free 14-day pilot, invite-only. Request access — no credit card, no sales call required to try it.