40 hours of lease audits a month, cut to six.

Where they started.

A mid-sized Westchester commercial portfolio was spending 40 hours a month across the operating PM and a paralegal on lease audits, CAM reconciliation run out of Excel, and an option-date calendar living in someone's Outlook. Yardi held the data; nothing useful came out of it without manual archaeology.

The growth math was broken: adding the next 20 leases — historically a one-person problem — was about to become a multi-hire problem, because the existing team was already underwater on the manual layer.

What we built.

We left Yardi in place as the system of record and built a custom layer on top: a thin extraction service landing lease, tenant, and ledger data in a dedicated warehouse, and a dashboard tuned to the decisions the portfolio actually makes.

At the intake stage, an AI lease-abstraction agent: field-by-field extraction with page citations and a deterministic validator, a side-by-side reviewer UI against the source PDF, and write-back to Yardi only on human approval. 96–98% field accuracy, with the residual caught in a 15–25 minute review.

Downstream agents watch the structured data: CAM variances surfaced with supporting line items, option windows alerted at 90, 60, and 30 days, critical dates re-checked whenever an amendment lands.

Where it landed.

Lease audit workload dropped from 40 hours a month to roughly six — an 85% reduction, with the remaining hours spent on judgment calls rather than data entry. The first quarter's CAM reconciliations alone caught $14,000 in under-billed recoveries.

One lease option that would have lapsed under the old Outlook-calendar process was exercised on time. The team redirected the recovered hours into proactive renewals — revenue work, not spreadsheet work.