• Real estate · CRE
  • Multi-agent AI
  • Web · API · MCP

Relm: multi-agent AI that finds hidden deal risk in commercial real estate

From an address to a sourced Intel Report and 10-year pro forma in minutes. Specialist AI agents research taxes, units, comps, permits and documents in parallel, and every number cites its source.

Minutes
address to Intel Report
44
MCP tools for AI agents
100%
findings source-cited
Relm Pro cover: AI-powered underwriting for real estate acquisition teams, beside a modern apartment buildingVisit live product (opens in a new tab)

01 · Challenge

CRE diligence takes analysts days of pulling records, and fatal issues like CO mismatches, liens or reassessments often surface only after the LOI.

02 · Approach

  • Built a multi-agent system: units, tax, comps, POI and pro-forma agents run in parallel and show their reasoning and confidence.
  • Added document intelligence with OCR, extraction and vector search over rent rolls, OMs and leases.
  • Exposed everything as a REST API and an MCP server, so firms can point their own AI tools at Relm.

03 · Results

  • Days of diligence compressed into minutes.
  • Every finding traceable to a source record.
  • Enterprise tenant isolation across database, vector store and AI context.
The product

Find hidden deal risk before you sign.

01

Intel Report

One search researches tax records, units, comps, demographics, listings and your documents in parallel.

02

Agent-built pro forma

A dedicated agent owns each line item and shows its reasoning, assumptions and confidence.

03

Chat analyst

Ask questions or request changes, each shown before and after, applied with one click and logged.

04

Document intelligence

OCR, extraction and vector search across rent rolls, OMs, P&Ls and leases.

05

API + MCP server

44 tools exposed over REST and MCP, so firms plug Relm into Claude, Cursor or their own agents.

06

Enterprise security

Tenant isolation across database, vector store and AI context, AES-256 encryption and an audit trail.

How it works

  1. STEP 01

    Address

    Enter any US property.

  2. STEP 02

    Research

    Units, tax, comps and POI agents run in parallel.

  3. STEP 03

    Model

    A 10-year pro forma builds itself, line by line.

  4. STEP 04

    Decide

    Review the flagged risks, then export to Excel or PDF.

Integrations

Wired into the systems that matter.

  • LLMs
  • Vector DB
  • OCR
  • MCP
  • REST API
  • Excel export
  • 20+ data feeds
What we delivered

Built and shipped for Relm.

  • Multi-agent AI
  • LLMs
  • RAG
  • Vector search
  • OCR
  • MCP server
  • REST API
Services used
AI Agent DevelopmentEnterprise LLM & RAG
Industry
Real Estate

Building something similar?

Start with a free AI audit. You get a fixed-price plan after a 2-week discovery sprint.

Book a free AI auditSee pricing
NEXT CASE · Fintech · Shared banking
Switch: shared banking for clubs, teams and friends