Main site

Real Estate AI agent · Expense recovery

Operating-Expense Reconciliation AI Agent

Build the expense pool from the ledger, cite the clause behind the line, hold the pack for the asset manager who issues a statement a tenant audits, not a price for a buyer.

Typical delivery
4–6 weeks
Deployment
Your stack
Manager issues
Pre-issuance
After launch
Agent Care

What this agent does

Builds the pack, not the decision to charge

In

  1. 01

    A cost is coded in the ledger, and the pool is drawn from the trial balance an auditor will later be shown.

  2. 02

    A line is tested against that lease clause, because the same account is recoverable next door and excluded here.

Reason

  1. 03

    A line no clause covers stays unclassified and is raised for a person, never defaulted into the recoverable pool.

  2. 04

    A gross-up runs on the variable components the clause names, and on the base year too where the lease has one.

  3. 05

    A cap is run on its own convention — compounded or not, rolling or base-year — and never on a house convention.

Decide

  1. 06

    A share is computed on the denominator the clause names; ANSI/BOMA Z65.1-2024 measures area, never the pool.

  2. 07

    A statement falls due on that lease's own date, and the objection, inspection and billing-bar clocks run apart.

Out

  1. 08

    A statement is not a tenant letter: this proves what may be charged, and a commercial lease has no approved form.

  2. 09

    Execute write actions only inside the approval boundaries agreed during implementation.

Product statement

The agent assembles the pool, the allocation and the pack; a named asset manager issues the statement, and the tenant tests it against the clause.

Example workflow

One tenant, ledger to statement pack

  1. 1Reconciliation year closesGeneral ledger, lease abstracts, estimate billings and the prior-year statement
  2. 2Pool assembledInclusions, exclusions, amortisations and gross-ups, each with its lease citation
  3. 3Pack draftedShare, estimate-to-actual bridge, unclassified lines and confidence
  4. 4Controls appliedClause-citation checks, cap and gross-up arithmetic, clock checks and confidence threshold

No human action required

Stages 1 to 4 run unaided, and nothing is issued or billed at any of them — the agent is assembling, and the asset manager's lane opens at the confidence gate.

5DecisionBranches at the confidence threshold
High confidence

Goes to the asset manager to approve.

Low confidence

Adds a lease-counsel read first.

Manager approval

The pack is held with its clause citations, its unclassified lines and the confidence.

Approve · Correct · Send to counsel review

Approved — released for issue

  1. 6Billing systems updatedOnly where write access and approval policy allow it
  2. 7Outcome evaluatedCitation coverage, corrections, on-time delivery and audit outcomes

Corrections

Every manager correction is counted in the evaluation.

What should not run autonomously

Human approval stays in control

Outside the boundary — human approval required

8 items

  • Issuing or certifying the reconciliation statement.
    • Deciding that a cost is recoverable under a clause.
    • Resolving or settling a tenant audit claim.
    • Signing the California attestation under § 1950.9.

  • Automation boundary

    Agent acts unaided

    • ✓Build the expense pool from the ledger, one coded line at a time.
    • ✓Carry the lease clause with the line it puts there.
    • ✓Compute the share, the cap and the bridge on the clause.
    • ✓Run the lease clocks and assemble the file behind an audit request.

    Any write happens inside the boundaries agreed at implementation, never ahead of the manager.

    • Re-billing a year already invoiced or closed.
    • Changing the allocation method for a tenant.
    • Granting an audit right the clause does not give.
  • Changes to clause, cap or gross-up rules.

Example output

One pool line, annotated

Everything the agent assembles is attached to the invoice and the clause behind it.

Reconciliation output · single tenant

Illustrative example
Tenant
Neighbourhood retail
Pool line
Management fee billed on the property, not the common area
Amount in pool
$41,860.00
Clause of record
Lease § 4.4.1
Confidence
84%
Who issues
Named asset manager

As received

Taken from the vendor invoice and the lease exclusion clause — nothing on this side is assumed by the agent.

Sources used

  • Vendor invoice on file
  • Lease exclusion clause
  • Prior-year statement

Why it is raised

Clear Lake Center (18 July 2013) held a fee like this limited to the common area.

What the score decides

Below the configured threshold the pack picks up a counsel read before the manager sees it.

Action

ApproveCorrectSend to counsel review

Value

Where AI adds value

The same four claims, placed at the point in the workflow where each one applies.

Where the value lands

Value 01 – 04

Every tenant

From the ledger and the lease

03Assembly

Work from the clause

Draw on the clause as written — reported case law on late billing and gross-ups is thinner than assumed.

01Approved path

The clause decides, not custom

Routine pools, shares and packs arrive already assembled.

02Human review

Send review to the contested lines

Uncited lines and unusual gross-ups are marked, so the counsel read starts where the exposure concentrates.

04

Build an evidence trail

The cost, the clause it was tested against and the manager who issued stay on the statement.

Integrations

Typical integrations

Five system groups connect to the same agent. Which of them are in scope is decided in discovery.

  • General ledger

    • Sage Intacct
    • NetSuite
    • Oracle
    • SAP exports

  • Property accounting

    • Yardi Voyager
    • MRI
    • RealPage
    • Entrata

  • Lease data

    • Lease abstract libraries
    • Clause and exclusion records

Agent

Reconciliation assembly

  • Reads the ledger
  • Cites the clause
  • Holds for issue

  • Tenant billing

    • Yardi CAM
    • MRI recovery
    • Billing and AR exports

  • Observability & evaluation

    • OpenTelemetry
    • Langfuse
    • Supported monitoring/evaluation sources

Integration availability depends on the client's existing systems and API access.

Agent controls

Six layers between the model and the statement

Six bands, each inside the last. Anything that survives them is drawn in the map below.

L6 · Outermost — last line of defence

Inward → L1 · closest to the model

L6Rollback / safe modePull assembly back to pool-only when evaluation or production signals degrade.Roll back
L5Version monitoringTrack model, prompt, clause-rule and property-configuration changes.Track
L4TraceabilityRecord the invoice, the clause, the allocation, the corrections and the approval time.Record
L3Manager approvalHold the pack for the named asset manager; it governs release, not whether a line is recoverable.Gate
L2Policy guardrailsTest each line against the clause set abstracted from that lease; a failure returns the pack.Restrict
L1Confidence thresholdsRoute low-confidence packs to a counsel read before the manager sees them.Require review
Model corePack produced — pool, exclusions, share, bridge, unclassified lines and confidence
L1 – L2
Test whether a line may stand
L3
Puts the release in the manager's hands
L4 – L5
Keep the cost and the clause behind it
L6
Narrows to pool assembly when signals degrade

How Nestack evaluates it

Evaluate the assembled pack — not only the total at the bottom.

Coverage runs the whole depth of the workflow, and every layer is cut by slice.

Surface — the statement the tenant receives

  1. E1Final-output evaluationDid each pool line match the invoice and the clause it came from?
  2. E2Step-level evaluationDid the agent use the right lease, cap convention and gross-up rule?
  3. E3Tool evaluationDid it read the correct tenant and the correct reconciliation year?
  4. E4Confidence calibrationDo low-confidence packs actually attract more manager corrections?
  5. E5Slice evaluationHow does performance change across specific lease and tenant types?
  6. E6Business outcomeHow many lines were corrected or contested after the statement went out?

Floor — the money the landlord gives back

Failure modes

Where each failure originates in the agent

Seven failure modes, set at the point the statement pack first breaks.

Agent lifecycle

Direction of processing →

  1. 01 · Retrieval

    1 mode

    • FS-03

      Stale trial balance

      A post-close reclassification lands after the pool was drawn.

    Stage gathers

    Ledger lines, lease clauses, estimates and areas

  2. 02 · Reasoning

    2 modes

    • FS-04

      Fee billed on the property

      A management fee is charged on the whole property, not the common area.

    • FS-06

      Cap run on house convention

      A non-compounded cap is run as compounding and widens each year.

    Stage proposes

    Pool, exclusions, share, bridge and confidence

  3. 03 · Tool / write

    2 modes

    • FS-02

      Unclassified line defaulted

      A line no clause covers is placed in the pool rather than raised.

    • FS-05

      Base year not grossed up

      Gross-up runs on the comparison year alone and compounds.

    Stage writes

    Only where write access and approval policy allow it

  4. 04 · Output

    1 mode

    • FS-01

      Statement lands late

      The pack goes out on the house calendar, not the lease date.

    Stage returns

    The statement the manager issues and the tenant tests

  5. 05 · Change / Version

    1 mode

    • FS-07

      Silent clause drift

      A rule change widens what the agent treats as recoverable.

    Stage tracks

    Model, prompt, clause rules and property config

  • Sev-1 · a charge leaves the boundary
  • Sev-2 · a wrong line reaches the manager
  • Sev-3 · source degrades, pack to review
See the complete Real Estate failure-mode catalogue

Affected slices

A property figure can hide one tenant

A property-level clause-citation figure can look sound while a few tenants absorb nearly all of the lines no clause covers. Nestack reports the uncited-line rate by slice, not only in total.

Slice performance — reported separately, not only in aggregate

Illustrative example
SliceFailure rateLiftLift vs. thresholdStatus
Base-year office leases6.8%3.7×Review
Capped controllable expenses4.6%2.5×Review
Ground-floor retail tenants2.9%1.6×Watch
Single-tenant net leases1.7%0.9×Normal

Bar: uncited-line-rate lift vs. single-tenant net lease baseline · scale 0–4.0× · tick marks the 2.0× review threshold

2 of 4 slices over threshold

Evidence-linked improvement

The season closes into the regression suite

The loop shuts when the disputed charge is a regression case. That suite is what the next statement issued is measured against.

Improvement cycle · five stages

Switchback — the path turns at Improve and returns at Learn

  1. 01

    Detect

    Uncited lines rise in one lease type.

  2. 02

    Diagnose

    The management fee that was billed on the property, not the common area, is read back until the cause narrows to one.

  3. 03

    Improve

    Any change ships numbered, with the statements that caused it attached.

  4. 04

    Verify

    One statement case still failing is enough to hold the release.

  5. 05

    Learn

    One case joins the suite, one line joins the recovery record.

  6. Learn → Detect

    The return edge. Next season's detection meets a suite one case longer.

Typical build scope

Twelve workstreams across six weeks

The build scope read against the delivery timeline. Week structure follows the six-week plan — discovery, sources, reconciliation workflow, evaluation, integration, then production validation and handover.

Workstream

Week 1Week 2Week 3Week 4Week 5Week 6

  1. 01Recovery workflow discovery and boundary setting.Week 1
  2. 02Ledger and lease source assessment.Week 1, Week 2
  3. 03Lease clause, exclusion and gross-up rule mapping.Week 1, Week 2
  4. 04Ledger extraction and pool assembly.Week 2
  5. 05Allocation, gross-up and cap arithmetic.Week 2, Week 3
  6. 06Confidence scoring and exception routing.Week 3
  7. 07Manager approval workflow.Week 3, Week 4
  8. 08Ledger and billing integration.Week 5
  9. 09Exclusion and gross-up cases.Week 4, Week 5
  10. 10Guardrails and issuance controls.Week 4
  11. 11Statement-trail instrumentation.Week 5, Week 6
  12. 12Deployment, documentation and Agent Care handover.Week 6

12 workstreams · 6 weeks · bar shows the weeks a workstream is active — several run in parallel

Final scope and sequence confirmed in discovery

Engagement tiers

What each tier includes

Rows are the capabilities named in each tier's scope. Higher tiers include everything below them.

Capability✓ in scope · — not at this tierPilotOne property, one seasonProductionProduction billing systemsAdvancedMultiple properties / states
Introduced at Pilot
Assembly to your clause set✓✓✓
Manager approval✓✓✓
Clause-citation baseline✓✓✓
Introduced at Production
Reporting by tenant—✓✓
Approval workflow in your systems—✓✓
Approved write-back—✓✓
Ledger-and-lease integration—✓✓
Introduced at Advanced
Multi-clause and multi-state rules——✓
Multi-stage landlord approvals——✓
Large tenant rosters——✓
Multi-property recovery controls——✓
Build priceFrom $5,000From $8,000Custom quote
Final build price
Confirmed after discovery based on integrations, workflow complexity, transaction volume, approval controls and deployment requirements.
Separate from build
Build pricing is separate from recurring Agent Care, which covers managed monitoring, evaluations, incidents and verified improvements after launch.

What we need from you

What you bring, and what we build with it

Each input maps to a piece of build scope and a week in the delivery timeline.

  1. 01Your lease abstracts and the clauses they carry

    Ledger and lease ingestion and clause binding

    Week 1

  2. 02The statement issued last year for each tenant

    Pool-assembly baseline and the uncited-line register

    Week 2

  3. 03Your cap, gross-up and amortisation conventions

    Lease clause, exclusion and gross-up rule mapping

    Week 1

  4. 04Access to relevant APIs, feeds or exports

    Ledger, lease and billing assessment, then integration setup

    Week 2

  5. 05Statements you would not want audited

    Audit cases and the evaluation suite

    Week 4

  6. 06What no statement may bury

    Confidence scoring, exception routing, guardrails and approval controls

    Week 3

  7. 07Named asset managers to review packs

    Manager approval workflow, then pilot and production validation

    Weeks 5–6

Nothing else is required

Deployment, documentation and Agent Care handover are ours.

Delivery timeline

Four phases across six weeks

The plan is measured in worked weeks and not in slide space, and so week five carries two bands.

Phase

W1

W2

W3

W4

W5

W6

  1. DiscoveryW1
  2. BuildW2 – W3
  3. EvaluateW4 – W5
  4. Pilot & LaunchW5 – W6

Week focus

W1Recovery workflow discovery, clause mapping and the automation boundary

W2Ledger and lease integration and the pool-assembly baseline

W3Reconciliation workflow, confidence logic and approval controls

W4Evaluation suite, clause-citation checks and failure-mode testing

W5Billing integration, pilot tenants and targeted corrections

W6One reconciliation season run under the asset manager, then Agent Care handover

Reading the band
A bar covers the weeks its work is named in, and nothing else. The week 5 overlap is real, not padding.
At the end of W6
Once the season clears, Agent Care takes over the running agent.
Duration
Six-week plan shown · typical delivery 4–6 weeks depending on scope confirmed in discovery.

Next step · Real Estate AI agent

Build a reconciliation agent around the clauses your own leases set.

Show us the leases, the ledger and the clocks each clause sets. Your asset manager is the person who signs the statement and then defends it to a tenant auditor holding the invoices. Underwriting reads an operating statement to price a building; this builds one a tenant can audit — and we found no error statistic for this workflow with a published sample, method or date, so the case is the contract, line by line.

Typical delivery
4–6 weeks
Deployment
Your leases and ledger
Manager issues
Pre-issuance
After launch
Agent Care available

Nestack Agents · Expense recovery

AGT-RE-20 · Agent Care available after launch

Build a reconciliation agent

Four questions so the call starts from your actual setup — then pick a time that suits you.

About you
Your portfolio

No obligation. We’ll map the workflow, name the automation boundary and tell you if an agent is the wrong answer.