BigDeal AI: a real estate CRM with an agent that asks before it acts
Leads, inventory, site visits, bookings and commissions in one system, with an agent that stops for a person before anything costly.
- AI agent
- In-house product
- Live
At a glance
Client BigDeal AI, an OutGrow product
Client
BigDeal AI, an OutGrow product
Industry Real estate sales
Industry
Real estate sales
Year 2026
Year
2026
Status Live
Status
Live
Every case on this site answers the same four questions, so you can compare them.
Before
Real estate teams run on portals, WhatsApp and spreadsheets. Leads go uncalled, two salespeople book the same flat, and payments get lost in a sheet.
What it does now
Leads arrive from Meta ads, portals, WhatsApp and the website, and duplicates are caught by phone number.
New leads are assigned in turn, and an alert fires when one goes too long without a call.
Inventory is tracked unit by unit, so a flat can't be booked twice.
Site visits, bookings, collections and channel-partner commissions sit on one record.
Ask the agent a question in plain language and it uses up to 30 tools to answer or act, in at most 8 steps.
Where a person decides
9 of the agent's 30 tools, such as recording a payment, sending a WhatsApp message or deleting a record, only create a proposal. An admin approves it before anything happens, including on scheduled runs at night.
Result
Tools the agent can use
At most 8 steps a run
Tools that wait for a person's approval
133 / 133
Tests passing
How we built it
Built the CRM core: leads, inventory, pipeline, site visits, bookings, collections and commissions.
Wrote the rules: duplicate checks, round-robin assignment, response-time alerts and a check every 15 minutes.
Built the agent loop and its 30 tools, with approval required on the 9 that cost money or reach a client.
Added the ad creative studio and a price-list importer.
Moved it to Cloudflare Workers with D1 for data.
Stack
- Next.js
- React
- TypeScript
- Cloudflare Workers
- Cloudflare D1
- LLM with tools
- Vitest
From the project
More work like this
Every inquiry read, scored and routed, at any hour
Our whole operation, readable and writable by any AI model
TeamBoard Studio: a whiteboard where the AI is a teammate
Your case study starts with an audit.
One to two weeks inside your operations. You get a map of where time and money leak, the return on each fix, and a fixed scope before we build anything.
Most businesses we meet have paid for an AI tool nobody uses. We start from your numbers, so the first build is chosen by payback.
Zain Khan and Anwar Sheikh
Founders, OutGrow
Outgrow
Intelligence
Studios.