Workflow and actors
The data team prepares closings, leads, contracts and incidents; AVM, scoring, churn and classification models generate signals. Valuation, sales and property-management teams act on them in their existing tools.
Service · Custom AI · Real estate & property
Custom AI for real estate that learns from your portfolio.
Valuation models trained on your real portfolio, deal-close probability scoring, structured extractors from listings and spec sheets, churn prediction and incident classification, trained on your data, not on portal averages.
The workflows, examples and figures on this page are illustrative composites and modelled targets, not measured client results. In a real project, we define the baseline, thresholds and human review with your data before rollout.
What we solve
Your portfolio doesn’t look like the market average.
Portal valuations and generic AVMs assume an average market. Your portfolio has a specific mix of typologies, areas, finish levels and target buyers, and the difference between closing in six weeks or six months sits exactly there. Same story with off-the-shelf CRM lead scoring: it treats a real buyer the same as a Sunday browser.
We build models trained on your history: an AVM tuned to your asset type, deal-close probability scoring per lead, structured extraction from listings and spec sheets, tenant-churn prediction and automatic incident classification by severity. Integrated with HubSpot, Salesforce or Pipedrive without replacing anything.
Implementation contract
How it runs in production
Systems and data
HubSpot, Salesforce or Pipedrive supplies commercial activity; portals and the PMS supply assets, comparables, closings, contracts, occupancy and incidents, with permissions and purpose separated by process.
Exceptions and risks
Areas with few comparables, unusual assets, incomplete data, market shifts, low-confidence leads and ambiguous emergencies are flagged; the AVM is not presented as an official mortgage appraisal.
Human review
The valuation lead approves outputs below the threshold or outside the agreed range, sales retains pricing and offers, and the property manager validates priority and supplier for critical incidents.
Implementation pattern
Separate versioned models for each decision, batch or real-time scoring, drift monitoring and traceability for features, confidence, review and final decision.
Relevant integration
Signals are written to fields in the existing CRM and PMS; HubSpot, Salesforce or Pipedrive connects to portals, portfolio and incident systems through APIs or an intermediate connector.
What we build for this sector
Use cases that ship to production.
Valuation
AVM tuned to your portfolio
Valuation model trained on your closed deals, real comparables and non-standard features (orientation, views, refurb quality). Outputs a confidence interval and the supporting comparables, not a single point estimate.
Modelled target: MAPE −25% vs generic AVM, by typology
Lead scoring
Deal-close probability scoring
Model predicting the probability a lead will close within 90 days, using portal behaviour, message history, sought typology and financing signals, not only fields the agent typed into the CRM.
Modelled target: Top 20% leads = 60–70% of closings
Extraction
Structured extractor for listings and spec sheets
Reads spec sheets, dimensioned plans and portal listings and returns structured JSON with surfaces, finishes, fittings and key differentiators, ready for CRM and for consistent commercial materials.
Modelled target: 120+ fields per asset
Predictive
Tenant-churn predictor
Model trained on historic move-outs, estimating non-renewal probability at 60 days by combining incidents, late payments, communication patterns and local market data, to enable a real retention window.
Modelled target: Recall > 0.75 on move-outs, by operator
Operations
Incident classifier by severity
Reads incoming incidents on any channel, classifies them by trade (plumbing, electrical, appliances…) and severity (critical, urgent, schedulable), and proposes SLA and supplier, ready to route to the property manager.
Modelled target: < 60 s to classify and route
Illustrative composite scenario · modelled figures and targets, not client results
A portfolio of 2,300 units under management.
Mixed operator: residential brokerage and build-to-rent property management in Madrid and Barcelona. Three years of history in HubSpot and a custom PMS. Illustrative composite scenario; the figures below are a modelled baseline and targets, not measured client results.
Modelled baseline
Baseline assumption: Listing prices are set using portal AVMs and agent judgement, 8–12% deviation from closing price. The commercial team works every CRM lead the same way; close rate around 4%. Tenant move-outs are spotted only when notice arrives, with no time to retain. Incidents are classified manually each morning.
Modelled target
Modelled operating target: The custom AVM cuts listing-price deviation to 3–4% and lets the team adjust portfolio pricing with data. Lead scoring focuses commercial effort on the top 20%, close rate climbs to 9%. The churn predictor flags 78% of move-outs 50–60 days ahead, opening a real retention window. Incidents arrive classified and routed to the right property manager.
Modelled target: +125% lead-to-closing conversion
We connect to your stack
Integrations that matter in this sector
CRM
HubSpot
Mid-market CRM with broad APIs, a natural fit for sales agents and lead enrichment.
CRM
Zoho CRM
CRM with strong adoption among Spanish SMBs, automations and agents at a contained cost.
COMMS
Microsoft 365 / Outlook
Email, calendar and SharePoint as channel and context, triage, drafting and RAG over your inbox and files.
Frequently asked
What clients ask us
- 01
Do we have enough representative data to train our own AVM?
It’s the first question we answer. We audit your history, number of closed deals, distribution by typology and area, quality of closing data, before promising any uplift. If the data isn’t there, we say so: sometimes the answer is to enrich with public and licensed sources, sometimes it’s starting with a simpler model and retraining as history grows. No fluff.
- 02
How does a custom AVM fit with regulation and official appraisals?
The model is an internal pricing, comparables and decision-support tool, it doesn’t replace a lender-approved appraisal for mortgage origination. We agree a confidence threshold and variation range: any valuation below the threshold or outside the range stays pending until the designated valuation lead approves it. We document the dataset, features, performance, review and final decision for every valuation. If you later want to move toward a regulated AVM, we start from the European framework for automated valuation models.
- 03
Do we need to change CRM or PMS?
No. We connect to HubSpot, Salesforce, Pipedrive, Idealista and the most common build-to-rent PMSs via API or intermediate connector. Predictions are written into properties of the existing CRM so the commercial team sees them in their usual flow, with no new tool to open.
- 04
What happens to the model when the market changes?
A model trained on a rising market stops being valid in a flat or falling one. So we set monthly retraining with new closings and continuous drift monitoring against actual market prices. If error rises above the threshold, the model is retrained or, if the magnitude justifies it, we revisit features and architecture. It’s part of the service, not an extra.
Same sector, other services
Trust
Safe, traceable AI,
enterprise-ready.
We design for privacy from the start, human control, traceability, usage limits, permissioning and documentation. For sensitive processes, we help assess risk and applicable obligations under GDPR and the EU AI Act.
- 01We never train models on your data without explicit authorization.
- 02Human review built-in for processes where risk demands it.
- 03Traceability: prompts, sources, permissions, errors and metrics, all documented.
- 04Privacy, security and control integrated from day one.
- 05Solutions engineered to be maintained, audited and improved over time.
GDPREU AI ActAEPDISO 27001 readyEU data residency
Personal diagnosis
We work with
few clients.
Every engagement is led personally by one of the partners. If there's a fit, you get a personal first read of your case within one business day, not a canned demo.
How we work
- 01Tell us which process eats your time
- 02Personal reply within one business day
- 0320-minute call, no demo, no pitch