Nestack Agent Care
Industries / Retail & E-commerce / Escheatment agent

Retail AI agent · Unclaimed property

Unclaimed-Property & Escheatment AI Agent

Pull the unclaimed balances into one population, age each item against the state rule that applies to it, and hand the officer who verifies the report the evidence behind each date.

4–6 weeksTypical delivery
Your stackDeployment
Per stateNamed officer
Agent CareAfter launch

What this agent does

Assembles the report, never verifies it

In
01

A credit goes unclaimed, and it is pulled into the population with the ledger it came from.

02

An item is matched to an owner, and the address of record decides which state has the first claim.

Reason
03

A dormancy period runs out under that state's rule, and the last owner activity sits behind the date.

04

A gift card is aged on its own state rule — California, New York, Texas and Delaware each answer differently.

05

A due-diligence letter falls due inside that state's own window and threshold, and the mailing is recorded.

Decide
06

An owner replies, and the item comes back off the report before anybody verifies it.

07

An exemption position is taken by a person, and the reason is recorded per item, per state.

Out
08

A Delaware notice lands, and its clock is diarised against the day it has to be answered.

09

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

Product statement

The agent assembles the report; in California a named officer of the private corporation verifies it before 1 November under Cal. CCP §1530(e), and no amount of assembly stands in for that.

Example workflow

One item, ledger to verification

AgentHuman
1Property record receivedLedger entry, card balance, payroll item or refund credit
2Owner and address resolvedAddress of record, priority state, owner identity and the activity behind it
3Item agedDormancy date, the state rule applied, its evidence and confidence
4Controls appliedPer-state dormancy rules, due-diligence windows, exemption reasons and confidence threshold
No human action required

Stages 1 to 4 run unaided, and nothing is verified or remitted at any of them — the agent is assembling, and the officer's lane opens at the confidence gate.

5DecisionBranches at the confidence threshold
High confidence

Goes to the verifying officer to read.

Low confidence

Adds an unclaimed-property counsel read first.

Officer review

The item is held with the rule it aged under, its evidence and the confidence.

Approve · Amend · Send to counsel
Approved — released to verify
6Property records updatedOnly where write access and approval policy allow it
7Outcome evaluatedPopulation completeness, ageing accuracy, letter reconciliation and post-filing corrections
Amendments

Officer amendments are counted in the evaluation.

What should not run autonomously

Human approval stays in control

Outside the boundary — human approval required8 items
Verifying the report a state requires of the holder.
Remitting property or cash to any state.
Deciding that an item has escheated.
Deciding an exemption or exclusion position.
Automation boundaryAgent acts unaided
Build the property population from the ledgers and the card platform.
Age each item against the state rule that applies to it.
Track the due-diligence window, threshold and retention per state.
Report what is in, what is out and the reason, before the verification.
Ageing is a computation against a rule. Verification is a named officer's act, not the agent's.
Answering the unclaimed-property question on a tax return.
Enrolling in a Delaware voluntary disclosure agreement.
Paying an owner after the dormancy period has run.
Changes to dormancy, exemption or approval rules.

Example output

One property item, annotated

What the agent records is attached to the ledger it was drawn from.

Escheatment output · single itemIllustrative example
Item
What the record shows
Priority state
Rule applied
Confidence
Owner contact
Merchandise credit
Issued in lieu of a cash refund and unredeemed since, with no owner activity behind it
New York
N.Y. Aband. Prop. Law §1315
91%
First-class letter sent, no reply
As receivedTaken from the ledger and the card platform as recorded — nothing on this side is decided by the agent.
Evidence behind the date Last owner activity Address of record Due-diligence letter
Why this rule appliesThe address of record sits in New York, so New York has the first claim on this item.
ActionApproveAmendSend to counsel
What the score decidesBelow the configured threshold the item adds a counsel read before it reaches the officer.

Value

Where AI adds value

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

Where the value landsValue 01 – 04
The ledgers in scopeFrom the source system
03Ageing

Age against the rule that applies

Draw on the last owner-generated activity and the dormancy rule of the state the address of record puts the item in. There is no federal statute here, and tax filing belongs to the accounting vertical.

01Approved path

The verification is the exposure

Delaware's Supreme Court held in Overstock, on 24 June 2020, that the absence of a record cannot ground a reverse false claim. The record that exists can.

02Human review

Where customer service ends

The returns agent refunds inside the deadline the payment type sets, and the payment-fraud agent works one transaction. This one starts where the balance becomes a state's property.

04Build an evidence trail

The item, the dormancy rule it aged under and the officer who verified stay on the report.

Integrations

Typical integrations

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

Ledgers and ERPSAP · Oracle
NetSuite · Microsoft Dynamics
Payroll and commissionsADP · Workday
UKG · Paycom
Gift card and stored valueBlackhawk · InComm
Givex · Card processors

Agent

Unclaimed property and escheatment

Reads the ledgers
Ages the items
Holds for the officer

Orders and refundsCommerce platforms
Returns and refund records
Observability & evaluationOpenTelemetry · 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 report

Six filters, tightest last. What none of the six catches appears in the map beneath it.

L6 · Outermost — last line of defenceInward → L1 · closest to the model
L6Rollback / safe modeHold the report at unverified when evaluation or production signals degrade.Roll back
L5Version monitoringTrack model, prompt, dormancy-rule and state-configuration changes.Track
L4TraceabilityRecord the item, the last-activity evidence, the letter and what the officer saw.Record
L3Officer reviewHold the pack for the named officer; it governs release, not whether the ageing is right.Gate
L2Policy guardrailsTest items against the per-state dormancy and due-diligence rules; a failure returns the item.Restrict
L1Confidence thresholdsRoute low-confidence items to a counsel read before the officer sees them.Require review
Model corePopulation assembled — items, dormancy dates, evidence, letters and confidence
L1 – L2Test whether an item may stand
L3Puts the release in an officer's hands
L4 – L5Keep the item and the rule behind it
L6Holds the report at unverified when signals degrade

How Nestack evaluates it

Evaluate the whole population — not only the finished report.

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

Surface — the report the state reads
Depth of coverage ▼
E1Final-output evaluationDid each item age against the rule of the state that applies to it?
E2Step-level evaluationDid the agent use the right ledgers, addresses and state configuration?
E3Tool evaluationDid it read and write the correct entity and the correct property record?
E4Confidence calibrationDo low-confidence items actually attract more officer amendments?
E5Slice evaluationHow does performance change across specific property types and states?
E6Business outcomeHow many items needed an amendment or a correction after the report went?
Floor — the outcome the officer answers for

Failure modes

Where each failure originates in the agent

Seven failure modes, set at the lifecycle stage each one begins in.

Agent lifecycleDirection of processing →
01 · Retrieval1 mode
GW-03

Processor activity field

A last-activity date is read from a field the processor also touches.

Stage gathersLedger items, card balances, addresses and state rules
02 · Reasoning2 modes
GW-04

Reset without evidence

A later purchase resets the clock on an unrelated credit.

GW-06

Exemption without a state

A counterparty exemption is carried into a state without one.

Stage proposesDormancy date, the rule applied and confidence
03 · Tool / write2 modes
GW-02

Letters on one national date

One mailing schedule is run at states whose windows differ.

GW-05

Reply lands after extract

An owner reply arrives after the report is pulled.

Stage writesOnly where write access and approval policy allow it
04 · Output1 mode
GW-01

Shipping address used

The address of record is displaced by a recent shipping address.

Stage returnsThe report the officer verifies and the state reads
05 · Change / Version1 mode
GW-07

Silent dormancy regression

A model or rule change shortens what is aged in.

Stage tracksModel, prompt, dormancy rules and state config
Sev-1 · agent acts outside the boundary Sev-2 · a wrong item reaches the report Sev-3 · source degrades, item routes to review

Affected slices

One property type can carry most of the gaps

A filing-level population-completeness figure can look sound while one property type absorbs most of the missed and misaged items. Nestack reports completeness by property type, not only in total.

Slice performance — reported separately, not only in aggregateIllustrative example
SliceFailure rateLift Lift vs. thresholdStatus
Gift cards and stored value7.7%3.7× Review
Merchandise and store credits5.5%2.6× Review
Accounts-payable credits3.2%1.5× Watch
Payroll and commissions1.2%0.6× Normal
Bar: population-gap lift vs. payroll-and-commissions baseline · scale 0–4.0× · tick marks the 2.0× review threshold 2 of 4 slices over threshold

Evidence-linked improvement

No cycle closes without a new test

A cycle closes when the misaged item is a regression case. That suite is what the next report verified is measured against.

Improvement cycle · five stagesSwitchback — the path turns at Improve and returns at Learn
01Detect

The population-gap rate rises in one property type.

02Diagnose

The gift card issued in one state to a customer who lived in another is read back through its ledger until the cause narrows to one.

03Improve

The fix carries a number, and the populations that forced it travel with it.

04Verify

Nothing goes out until every touched population case has passed again.

05Learn

It is kept permanently, and the dormancy rules are amended alongside it.

Learn → DetectThe return edge. Next time detection is measured against a suite that has grown by one.

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, ageing workflow, evaluation, integration, then production validation and handover.

Workstream Week 1Week 2Week 3Week 4Week 5Week 6
01Property population discovery and boundary definition.
02Ledger and card-platform assessment.
03Per-state dormancy and due-diligence rule mapping.
04Property-record ingestion and matching.
05Ageing logic and evidence binding.
06Confidence scoring and exception routing.
07Officer review workflow.
08Ledger and card-platform integration.
09Ageing and due-diligence cases.
10Guardrails and verification controls.
11Property-trail instrumentation.
12Deployment, documentation and Agent Care handover.
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 tier PilotOne entity, one state ProductionProduction ledger systems AdvancedMultiple entities / states
Introduced at Pilot
Ageing to your states and rules
Officer review
Population-completeness baseline
Introduced at Production
Reporting by property type
Approval workflow in your systems
Approved write-back
Ledger-and-card integration
Introduced at Advanced
Multi-entity rule configuration
Multi-stage finance approvals
High property volume
Multi-state dormancy controls
Build price From $5,000 From $8,000 Custom quote
Final build priceConfirmed after discovery based on integrations, workflow complexity, transaction volume, approval controls and deployment requirements.
Separate from buildBuild 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.

You bringWe build with it
01Your ledgers and the entities in scope Property-record ingestion and owner matchingWeek 1
02Representative unclaimed balances Completeness baseline, ageing and evidence bindingWeek 2
03Your per-state dormancy and exemption rules Per-state dormancy and due-diligence rule mappingWeek 1
04Access to relevant APIs, feeds or exports Ledger and card-platform assessment, then integration setupWeek 2
05Reports you would not want examined Ageing cases and the evaluation suiteWeek 4
06What no verified report may guess Confidence scoring, exception routing, guardrails and approval controlsWeek 3
07A named officer to review the pack Officer review workflow, then pilot and production validationWeeks 5–6
Nothing else is required Deployment, documentation and Agent Care handover are ours.

Delivery timeline

Four phases across six weeks

Weeks the job actually consumes decide each width drawn here, so two bands sit inside the fifth.

Phase W1W2W3W4W5W6
Discovery W1
Build W2 – W3
Evaluate W4 – W5
Pilot & Launch W5 – W6
Week focus W1Population discovery, rule mapping and the automation boundary W2Ledger and card-platform integration and the completeness baseline W3Ageing workflow, confidence logic and approval controls W4Evaluation suite, due-diligence checks and failure-mode testing W5Refund and payroll integration, pilot populations and targeted corrections W6One filing season run under the verifying officer, then Agent Care handover
Reading the bandEach bar sits on the weeks its work is named in, and no others. The week 5 overlap is real, not padding.
At the end of W6The final checks clear on live populations and monitoring moves to Agent Care.
DurationSix-week plan shown · typical delivery 4–6 weeks depending on scope confirmed in discovery.

Next step · Retail AI agent

Build an escheatment agent around the officer who verifies the report.

Show us your ledgers, your card platform and the officer who verifies. Not the filing you missed. The one you verified: Delaware's Supreme Court held in Overstock, on 24 June 2020, that an absence of a record cannot ground a reverse false claim.

Nestack Agents · Unclaimed property and escheatmentAGT-RT-19 · Agent Care available after launch