Illustrative engagements

Illustrative · BC investor

Weekly market intelligence instead of a stale PDF

The question: A British Columbia-based investor holding multi-family and mixed-use assets across three metros was relying on quarterly broker PDFs that were often outdated by the time they circulated. Leadership wanted a living read — absorption, rent trajectory and competitive supply — refreshed on a fixed weekly cadence.

What we delivered: A market intelligence dashboard pulling MLS transaction feeds, rental listing snapshots and municipal permit data into automated data pipelines. Predictive analytics models produced demand and price forecast bands with documented confidence intervals. Natural language processing summarized local news and policy changes into a Monday briefing folio.

Uncertainty & review: Models were back-tested over eighteen months of held-out data; largest errors occurred during rapid policy shifts — noted explicitly in the dashboard footer. A senior analyst reviewed every weekly read before publication. The investor used the intelligence to time disposition conversations — the decision stayed with their investment committee, not our studio.

Illustrative · Canadian REIT

Four hundred units, zero single view

The question: A Canadian REIT with roughly four hundred residential and retail units across Western Canada had rent rolls in three formats, operating data in an ERP and market comps in spreadsheets. Asset managers could not answer a simple question — what is our weighted occupancy trend by region? — without two days of manual work.

What we delivered: Portfolio analytics unifying asset records into one dashboard with drill-down by property, vintage and geography. Automated valuation model estimates sat beside actual NOI with clear labelling: model-generated figures, not appraisals. Scenario modules tested refinance and hold paths under rent-growth assumptions the client supplied.

Uncertainty & review: AVM outputs diverged from recent transactions in thin markets — flagged automatically when error exceeded a defined threshold. Quarterly retainer included model evaluation refresh and a human analyst walkthrough with the asset team. Outputs informed board materials; they were not investment advice.

Illustrative · Vancouver-area developer

Choosing between two land parcels

The question: A developer evaluating two candidate parcels in the Vancouver region needed a location-demand study before committing earnest money. Both sites looked plausible on a map; neither had a disciplined read of nearby supply pipeline, transit adjacency or demographic trajectory.

What we delivered: Geospatial location analytics comparing walkability proxies, employment node proximity, historical absorption in the submarket and permitted competing supply. Comparable analysis sets anchored asking-price context. A written intelligence brief ranked the parcels against the client's product criteria — with explicit caveats where census data was dated.

Uncertainty & review: Forecast ranges overlapped between the two sites for certain unit types, which we stated plainly rather than forcing a false winner. The developer's land committee made the call; our read narrowed the debate from gut feel to evidence.

Illustrative · Ontario brokerage

Lead scoring that respects the agent relationship

The question: A mid-size Ontario brokerage had leads sitting in a CRM inbox until they went cold. They wanted lead scoring based on engagement signals and property-interest patterns — without automating outbound messages that would feel robotic to clients.

What we delivered: A lead-scoring model trained on historical conversion patterns, integrated with their CRM via a secure data pipeline. Scores surfaced in the agent dashboard with explainability notes — which signals moved the number. Document AI extracted key fields from buyer preference forms to enrich features.

Uncertainty & review: Model performance varied by agent team; we recommended per-team recalibration rather than a single national threshold. Brokers reviewed scores before prioritizing follow-up. We are not a licensed brokerage; the tool supported their agents, not replaced them.

Illustrative · Property manager

Automated reporting for a scattered portfolio

The question: A property management firm overseeing commercial and residential assets needed monthly owner reports assembled from six source systems. Analysts spent the first week of each month copying figures into templates.

What we delivered: Automated reporting pipelines and an executive dashboard with occupancy, arrears, capex tracking and market rent positioning. Retrieval-augmented generation drafted narrative sections from structured data; an analyst edited every report before distribution.

Uncertainty & review: OCR on scanned invoices required a human validation queue for low-confidence extractions. Reporting cadence moved from manual to automated, but judgment on owner communications remained with the property manager's team.