Santo Terrace & Canyon View

A multi-billion MHPI portfolio is not run on spreadsheets. Each project is a bond-financed SPE with a lockbox waterfall, trustee-held capital repair & replacement reserves, recapitalization requirements, and (likely) periodic third-party capital needs assessments. The gap is probably not forecasting — it's that the forecasts are static PDFs, siloed per deal entity, and carry no lease-term or cross-portfolio lens.

Structural insight: cash generally can't move between project entities. "Where should we invest across the company" must respect per-project waterfall capacity — condition data alone gives the wrong answer.

• KEYWhat did the investment bankers ask for that was hard to produce? (This question finds the product.)
• OPENLast capital needs assessment per project — who, when, format, can we get the underlying models?
• OPENComponent-level inventories in the CNAs, or building-level allowances?
• OPENHow many project entities, which communities in which, reserve balances & waterfall position per project?
• OPENActual ground lease expirations per project; which extensions are live?
• OPENWhich Yardi modules are deployed — Fixed Assets? Facility Manager? Where do work orders live, and do they capture asset detail?
• OPENHow is next year's capex budget actually built today — inputs, owners, approval, lock date?
• OPENWhat flows to DoW/trustee already — and which metrics does leadership actually read?

Full 22-question diligence list in the Discovery Brief. Pitch shift: not "I'll build forecasting" — "your forecasts are trapped in five-year-old PDFs and per-deal silos; I'll make them living, comparable, and lease-aware."

Forecasting answers capital budgeting from age distributions; a census answers dispatch. A census that is 80% maintained is not a census — it's untrustworthy data. So the census is a bonus layer that fills in opportunistically; the forecast never depends on it.

Portfolio confidence share becomes a headline metric — "our 10-yr number is 40% verified / 45% logged / 15% inferred" is a sentence a banker can work with.

The bulk-purchase reality: they buy a dozen disposals into a shed; the PO never says which unit each landed in. So capture happens in two stages — procurement → inventory, then inventory → unit at install. The install log rides on the stock-pull they already must do.

• OPENIs shed/warehouse inventory tracked anywhere today, or is stage 1 net-new?
• OPENLabel spec & vendor (polyester/vinyl, heat+moisture rated) — and who prints?
• OPENTag deployment: normal maintenance sweep vs. one-time contracted pass?
• OPENWhat system closes work orders today — can install logging auto-close them?

Position: sit above it, don't compete. Tailorbird is data acquisition; this platform is the decision layer. Their output would land as one inferred-tier feed. An unproven vendor is a reason to have an independent decision layer.

• KEYAPI or bulk export — can the full asset table with confidence scores land in our Postgres, or is data locked in their UI?
• OPENPer-asset confidence scores exposed? Accuracy stats vs. later-verified ground truth?
• OPENError direction — do they miss assets or invent them?
• OPENExact Yardi requirements — modules, licensing tier, read-only?
• OPENData ownership at contract end — does the inventory leave with LMH?
• OPENRefresh cycle — snapshot or continuous re-inference?
• OPENPortfolios completed at 30K+ units — reference client?
• OPENMHPI experience — base access, DoW data handling?
• OPENStrong vs. weak asset classes (aerial-visible vs. inside-the-unit)?
• OPENGaps: flagged honestly, or filled with portfolio averages?
• OPENCost + timeline of last comparable engagement?

Santo doubles as the benchmark: if Tailorbird runs it too, compare their inferred inventory against ours, asset by asset.

Their goal: district personnel click any property and immediately know who owns what and who to call. Per property, track:

• TRACKElectric / gas / water / sewer provider (service territories are largely public GIS data)
• TRACKPoint of demarcation — where public utility responsibility ends and LMH's begins
• TRACKOn-base wrinkle: privatized utility owners (separate companies from housing privatization) — nobody has this in one map
• TRACKOutage contact + account numbers per service
• TRACKWho owns lines inside the community boundary (LMH vs. utility vs. base)

Santo (off-base, city land): likely SDG&E electric+gas, City of San Diego water/sewer — to verify, which is itself the demo of the workflow.

Cost reality: interface/API licensing is unpublished, negotiated, typically metered per property per month, and gated by entitlements. At 200 communities: potentially thousands to tens of thousands monthly. But geography changes a few times a year — a static hierarchy doesn't justify a live licensed sync. One-time bulk export (even CSV) and done.

• OPENWhich Yardi product/tier do they run — Voyager, Breeze, Breeze Premier?
• OPENCan Mike pull a unit directory / rent-roll export for Santo (resident columns stripped)?
• OPENAre appliance replacements logged as work orders — extractable history?
• OPENRevisit live sync only if Phase 4 (operational overlay) gets funded.