Lease Administration
Retail lease abstraction, anchored to the clause it came from.
Purpose-built for retail lease complexity — and for the thousands of retailers still carrying the hidden cost of a rushed ASC 842 migration.
Between 2019 and 2022 most retailers rushed an ASC 842 migration. To hit the deadline, implementation teams consolidated covenant data into a single free-text field instead of the purpose-built fields the new platform provided. Co-tenancy tracking, renewal alerts, CAM reconciliation — functionally out of reach ever since, and every lease added since compounds it.
The data is there. The system just can't use it.
The full argument, and why a general-purpose model didn't fix it
Generic AI wasn't built for retail leases.
Most abstraction tools were designed for standard commercial leases and reach 80–85% accuracy. For retail, that missing 15–20% is usually the most valuable part: the co-tenancy clause that could cut rent by 30%, the kickout provision on an underperforming store, the CAM audit rights nobody knew to enforce.
Surfaice was trained on retail lease complexity — percentage rent structures, co-tenancy provisions, exclusivity clauses, radius restrictions, and the negotiated carve-outs that vary deal to deal. Every extraction is anchored to an exact source clause. Where one cannot be found, it is flagged rather than invented.
Generic AI tools
83%
Average accuracy on retail leases
- Misses co-tenancy clauses
- Generic commercial training
- No clause-level citations
- Hallucination risk
Surfaice
97%+
With human review on every flag
- Retail-specific training
- Co-tenancy and kickout focus
- Exact clause citations
- Flags gaps, never invents
How it works
How re-abstraction works.
Your team approves every change before anything is written or deleted. Nothing moves without an explicit sign-off.
- 01
Upload the lease and your current system export
Surfaice reads the original lease PDF alongside a data export from your lease admin platform — Lucernex, MRI, SAP, Oracle or Procore. Processed in memory, deleted after the session.
Works for new leases and existing ones. New leases abstract directly into the correct fields; existing ones run a side-by-side comparison against what your system already holds.
- 02
Field-by-field comparison with confidence scoring
Every covenant is extracted and compared against what is in your system. High-confidence matches queue for batch approval. Conflicts, gaps and low-confidence extractions are flagged for individual review, with the source citation visible.
- 03
Your team reviews and approves
Batch-approve high-confidence data in one action. Conflicts and flagged items go to individual review with the source clause beside the decision.
Surfaice can remove the old, badly structured data as it writes the corrected data into the right fields — restoring your platform's functionality in a single pass. Nothing is deleted without explicit approval.
- 04
Written to your system of record
Approved data goes into the correct fields of your existing platform, not a catch-all notes field. Your system can then generate renewal alerts, flag co-tenancy triggers, run CAM reconciliation and produce portfolio analytics it could not before.
| Covenant | In your system now | Surfaice extraction | Status |
|---|---|---|---|
| Renewal option | See covenants field | 2 × 5-yr options; notice by Dec 31, 2026 §4.2 | Auto-approve |
| Co-tenancy clause | Not abstracted | Anchor occ. ≥ 60%; rent reduces 30% if below §11.4 | Auto-approve |
| Kickout clause | Not abstracted | Sales threshold $850K/yr; 90-day notice after Yr 3 §14.1 | Auto-approve |
| CAM cap | 5% (legacy field) | 5% cumulative (not compounded) §8.3 | Review — wording differs |
| Percentage rent | See covenants field | 6% of gross sales > $1.2M natural breakpoint §7.1 | Auto-approve |
| Radius restriction | 3 miles (legacy) | 5-mile radius §9.2 | Review — conflict |
Every extraction links to the exact clause in the source PDF.
What badly abstracted lease data costs you.
These are not hypothetical. They are operational failures happening across retail portfolios every quarter.
Kickout clauses you cannot trigger
Kickout provisions let you exit underperforming locations. If the clause is not properly abstracted there is no alert when sales hit the threshold, and you keep paying rent on a store you could legally walk away from.
Renewal deadlines missed by a day
Notice windows for renewal options and rights of first refusal are strict. Miss by a single day and the right is gone. Badly structured data cannot generate the alert — someone has to know to look.
CAM overcharges you are not disputing
Without easy access to your caps, exclusions and audit rights, pushing back on a landlord statement takes days. Most teams do not, and the overcharge recurs every year.
Co-tenancy reductions that never trigger
You negotiated a 30% rent reduction if anchor occupancy drops below 60%. If that clause lives in a free-text field there is no trigger: the anchor leaves and your rent does not change.
Portfolio analytics built on bad data
Every multi-site report, renewal strategy and lease-versus-buy decision runs on your lease data. If that data is unstructured, incomplete or wrong, so are the decisions.
35×
faster
Than manual abstraction — 12 seconds per lease against 4 to 8 hours.
<$10
compute cost
Per lease, against roughly $350 for legal or paralegal review.
100%
source-cited
Every data point linked to the exact clause it came from.
Lease abstraction, answered.
What does Surfaice extract from a commercial lease?
Critical dates, renewal options, rent steps, CAM terms, co-tenancy and kickout clauses, exclusive use, radius restrictions, insurance requirements and landlord work letter references — each tied to a clause citation so your team can verify without re-reading a 90-page PDF.
How is this different from manual abstraction or generic AI?
Extraction is deterministic and playbook-driven for the retail lease fields your portfolio team actually uses. Outputs map to Lucernex, Excel and Procore formats rather than unstructured summaries, and flagged fields route to human review when wording conflicts with the legacy abstract.
Which document types are supported?
NNN, gross and modified gross leases, letters of intent, amendments and work letters. The agent is tuned for multi-location retail portfolios where inconsistent abstract quality creates CAM, renewal and reporting risk.
Can we export abstracts into our lease administration system?
Yes. Structured exports include Excel tables, PDF reports, JSON data and Procore import formats, so abstracted fields feed lease admin, accounting and development workflows without re-keying.
How does clause risk review work?
Risk review flags non-standard language in termination, default, force majeure and liability sections. High-impact deviations surface with citations so legal and real estate teams can prioritise review before a renewal or disposition decision.
Fix your lease data. Get your system working again.
Bring one lease and your current export. You will see the difference field by field, with the clause beside each one.