From spreadsheets to one portal: AI matching, valuations and document tracking for a UK property brokerage
A UK property deal-sourcing brokerage ran every property, contact and deal from Google Sheets. We replaced them with one internal Airtable portal: AI matching, AI-written teasers, a deal pipeline, document and due-diligence tracking, AI valuations and an AI assistant that can see the whole database. It was handed over on 22 September 2026.
- Sheets migrated
- AI matching
- Deal pipeline
- AI valuation
Snapshot
- Client type
- Property deal-sourcing brokerage
- Country
- United Kingdom
- Industry
- Real estate: residential, commercial, and land and development
- System
- AI Automations, with an internal AI assistant
- Use cases
- Operations database · matchmaking · AI teasers · deal pipeline · document tracking · AI valuation · AI database assistant
- Stack
- Airtable (Team plan) · Make.com
- Timeline
- Single build from a proven template · handed over 22 September 2026
- Headline numbers
- No performance figures yet · matching switched on once buyer criteria are entered
What did the client need?
One system for every property, contact and deal, with automatic matching, visible paperwork and fast valuations.What did we build?
The finished version of a platform we'd refined for another UK brokerage, built in one go and adapted to this team's process.How does it run day to day?
Criteria match instantly, new listings match each weekday at 9:00 UK, and the team sends teasers and moves deals by hand.
What did the client need?
The brokerage lists properties it owns or controls, plus properties brought by agents who must show they hold the owner's authority to sell. Each deal needs its own letter of intent (LOI), proof of funds and fee agreements, and contacts and properties need NDAs and due diligence.
All of it lived in separate Google Sheets tabs. Nobody could see at a glance which buyers suited a new property, which deals were waiting on an LOI or proof of funds, which agents had shown authority to sell, or whether an asking price was realistic. Matching meant reading criteria by eye, and paperwork was tracked from memory.
What did we build?
We'd already built a property matching platform for another UK brokerage over many months. This client got the finished version in a single build, plus its own additions.
- Statistics dashboard: deals by stage, properties by type and source, contacts, total portfolio value, and combined buyer demand by property type and location. Every number clicks through to the records behind it.
- Team dashboard: contacts, residential, commercial and land stock, deals, every buyer's criteria in one list, and a management page ready for when the team grows.
- Matching: new buyer criteria are matched against all available stock at once; each criteria set is matched separately, so one buyer can want different things in different countries. Any property can also be matched to a chosen buyer by hand.
- Deals pipeline: from new match through follow-up, LOI, proof of funds, offer and completion to payment received. Every match becomes a deal, marked as AI or manual.
- Verification and due diligence: document statuses on contacts, properties and deals that update automatically when a file is uploaded.
- AI valuations: one click runs an agent that reads the property's data, researches the market online and returns a price range, a pricing verdict and its reasoning.
- AI assistant: a chat inside the portal that can see the whole database, answer questions, write PDF reports and save them to a property's record.
The data was migrated across from Google Sheets, and everything was built in the client's own Airtable workspace and Make.com account.
How does it run day to day?
Buyer criteria added ─► instant matching ─┐ Property set to Available ─► 9:00 weekday batch ─┼─► deal + teaser ─► team sends teaser ─► pipeline Document uploaded ─► status updated "Run agent" ─► AI valuation ─► review
- Only verified buyers are matched. The team sets a buyer to Active by hand once they've checked the buyer's criteria and verification.
- Only available stock is matched. Properties under negotiation, on hold, sold or blocked drop out of matching automatically.
- Teasers are sent by the team. Each new deal shows its teaser, and the team sends it, at the client's request.
- Valuations flag missing data. If required fields are empty, the agent says it can't run rather than guessing.
- Internal only, for now. External logins, referrals, team management and in-portal messaging are built but switched off, ready for a later phase.
What were the results?
There are no performance figures yet. The portal was handed over on 22 September 2026, and automated matching is switched on once the brokerage has entered its buyers' criteria. What was delivered is a single system replacing the team's spreadsheets, with matching, document tracking, valuations and an AI assistant built in.
- 22 Sep 2026Portal handed overDelivery date
- 1 buildFrom a proven template, not built in phasesDelivery
- Use cases in one portalDelivery scope
What did we learn, and what are the limits?
- A proven template makes a fast, complete build. Months of refinement on the first platform meant this one started with the right structure, pipeline and automations.
- Pipeline stages must follow the client's real order. We expected the LOI and proof of funds in either order. This client always gets the LOI first and needs proof of funds before an offer, so it got a dedicated stage.
- Walk the client through every stage. Two recorded walkthroughs surfaced changes a written spec wouldn't have, and became the team's training material.
- Only automate statuses with a document behind them. Terms sent or fee agreements spread across several documents stay manual, so the system never marks something done when it isn't.