Real Estate
Every portfolio, ready for a decision.
A real estate investment firm with more than 120,000 assets under management receives portfolios full of information, but poorly enriched. We built an AI platform on Azure that completes every asset, helps analysts prioritize and prepares analyses aligned with what each investor is looking for. Today each analyst prepares three times as many investment studies.
Discover the case- Incomplete assets arrive
- Their information is completed
- The most promising ones stand out
01
The starting point
A lot of information. Little that is useful.
The firm receives portfolios from very diverse sources, from a few hundred to tens of thousands of assets. They arrive with plenty of information, but spread across different formats and full of gaps: each asset lacks the data that is really needed to value it.
+120,000
Assets under management
Across Spain
+40
Pages per transaction
On average, between deeds, registry extracts and charges
4
Input formats
Excel, PDF, Word and email
This is how an asset arrived, and what it was missing.
The analyst obtained every missing data point by hand: searching the cadastre, estimating an approximate valuation, checking the market and reading the legal documentation.
Asset record
Home in Valencia
Arrives spread across Excel, PDF, Word and email
5loose pieces that arrive
Address and typeValencia, 46022. HomeExcel
Asking price€142,000Excel
Land registry extractScanned, unreadPDF
Seller’s report12 pagesWord
Attachments and photosUnclassifiedEmail
5missing data points, and how they were obtained
Cadastral reference and floor areaLook up the asset in the cadastre
Asset valueEstimate an approximate valuation
Sale and rental prices in the areaCheck property portals and market studies
Charges, occupancy and litigationRead the 40+ pages of the file
Fit with each investorMatch it by hand against what each one wants
02
The challenge
Time went into preparing, not investing.
Before valuing a single asset, the analyst had to complete it: unify formats, search the cadastre, estimate its value, check the market, read the legal documentation and, finally, tailor the analysis to each investor. Most of the week was gone before any decision could be made.
An analyst’s week
01
No two portfolios are alike
Each source names, orders and delivers the data its own way, in four different formats.
02
What is missing outweighs what is there
Without floor area, value or local prices, an asset cannot be compared.
03
Each investor wants something different
Area, type, ticket size, yield and risk change from one to another.
The challenge was not having more data, but turning it into timely decisions.
03
The solution
Information completes itself. The analyst decides.
We built the complete journey of an asset on Azure and its ecosystem: information is enriched automatically, the analyst prioritizes and validates, and the sales team presents each investor with what they are looking for. This is how the same home in Valencia goes through it.
Asset record
Home in Valencia
Presented to the investor by the sales team
- Portfolio4 formats
- Cadastre3 data points
- Valuation1 data point
- Market2 data points
- History1 data point
- Legal4 data points
Asset
Home, €142,000
AddressValencia, 46022
Type and priceHome, €142,000
Format4 formats, not unifiedLoadCommon record
Cadastre, value and market
94 m², valued at €168,500
Cadastral referenceCadastreIdentified
Floor areaCadastre94 m²
Year builtCadastre1987
Appraised valueValuation€168,500
Sale price in the areaMarket€2,140/m²
Rent in the areaMarket€11.8/m²
Similar transactionsHistory14 in two years
Legal status
Vacant, no litigation
ChargesNot reviewedFileCancellable mortgage
OccupancyNot reviewedFileVacant
Litigation and easementsNot reviewedFileNone
Documentation46 pages unread46 pages read and linked
Prioritization and review
Validated by the analyst
Investment potentialModelHigh
Risk levelModelLow
Estimated gross yieldModel6.9%
Position in the portfolio4 of 214 prioritized
Analyst reviewValidated
Next stepPassed to the sales team
Fit with investors
Fits 2 of 3 investors
Residential, eastern SpainFits
Retail units in large citiesDoes not fit
Discounted portfoliosFits
Investment studyPrepared for 2 investors
PresentationPresented to the investor
01Enrich
Load layer. Four formats, a single record.
The portfolio comes in as it arrives: Excel, PDF, Word or email. Each column is recognized even when every source names it differently, duplicates are removed and each asset is translated into a common record.
Automatic loading
With Azure Data Factory and Azure AI Document Intelligence
01Enrich
Enrichment layer. Every data point, from its source.
The record is completed with the cadastre, valuation systems, two sale and rental market sources and the firm’s own transaction history. What the analyst used to look up by hand now arrives on its own. Average portfolio completeness rises from 31% to 94%.
Automatic enrichment
With Azure Functions and Azure Data Lake Storage
01Enrich
Legal layer. Between 30 and 150 pages, read in minutes.
Deeds, land registry extracts and charge certificates are read, and only what is key for investing is extracted: charges, occupancy, litigation and easements. Each data point stays linked to the page it comes from.
Automatic reading
With Azure OpenAI Service and Azure AI Search
02Prioritize and validate
Prioritization layer. A score for every asset.
A model combines investment potential, risk level and projected yield. The portfolio sorts itself and every position has a reason.
Proposed by the model
With Azure Machine Learning
02Prioritize and validate
The analyst reviews everything and decides.
They review the complete record and the score, and validate or discard. If they validate, the asset goes to the sales team; if they discard it, the reason goes back to the model. For any question, they query the whole database in natural language through a dedicated agent.
Decided by the analyst
With Power BI and Copilot Studio
03Align and present
Investment studies. What each investor is looking for.
Each validated asset is matched against each investor’s criteria (area, type, ticket size, yield and risk) and a tailored study is prepared, ready for the analyst to review.
Automatic preparation
With Azure OpenAI Service
03Align and present
Sales presentation. Every proposal, with follow-up.
The sales team sends and presents each proposal, with its status and owner in the CRM. The investor’s response goes back to the model and refines the next prioritization.
Presented by the sales team
With Power Automate and CRM integration
Technology partner
We are a Microsoft Solutions Partner for Data and AI
A designation that recognizes our technical capability, experience and results building data and artificial intelligence solutions on Azure and the Microsoft ecosystem, for companies in consumer goods, industry, banking, pharma, retail and real estate investment.
04
The result
More time to invest, less to prepare.
60%
Less management time
Out of every 10 hours of preparation, 4 remain today
50%
More productivity
3 portfolios analyzed where 2 used to be
30%
Lower operating costs
Above all in manual loading and document review
x3
Investment studies
Each analyst prepares an average of 12 investment studies a month, compared with 4 before.
With the information already complete, time goes into analyzing and tailoring each study to what the investor is looking for.
4 days
To respond to a portfolio
Before, about three weeks
+25%
Proposals with investor interest
Because they are aligned with what each one wants
An analyst’s week, turned around.
Preparing data drops from 70% to 25% of the week. The time freed up goes to what only a person can do: validating, analyzing and preparing proposals for each investor.
05
What comes next
From analyzing portfolios to finding opportunities.
With every asset complete and every investor well known, the next step is to turn the process around: the platform proposes before anyone asks.
In production
Enrich, prioritize and align
Every portfolio received, completed and turned into tailored studies.
Next step
Continuous investor matching
Each new asset is instantly matched against what each investor is looking for.
Horizon
Active market search
Spotting opportunities beyond the portfolios received, with the same model.
How much value is in the data you already have?
Write to us and we will show you how to apply it to your portfolios.