The public record contradicts itself

The AI investigation engine for public spending.

Reckton compares what the city ledger, state registry, and IRS say about the same organization. When the records disagree, for example when a suspended vendor is still paid or filings conflict under penalty of perjury, Reckton shows the issue as evidence your team can re-verify. Built for audit firms and oversight offices.

Check one vendor See the San Francisco case

The proof

Follow the organization back to the record.

A real case structure showing the review condition, dollars, primary sources, uncertainty, and query needed to reproduce the comparison.

SF-BH-2026 · specimen extractA CERTIFIED

Six departments paid a vendor $2.87M in the fiscal year state records show its corporate status suspended

The issue

The city's published ledger records $2,868,481.87 paid to a behavioral-health vendor across 55 vouchers by six departments in fiscal year 2024. State records show a Franchise Tax Board suspension effective April 2, 2024, still in place at retrieval, and a charity-registry dissolution waiver. The published ledger carries no payment dates, so the amount paid on or after the suspension date is not yet established; that is the first record to pull. The overlap warrants review.

Dollar figure: $2,868,481.87 in fiscal year 2024, as recorded in the city's published vendor payment ledger (55 vouchers, six departments)
Entity match: State records matched on entity number and FEIN; the city ledger lists the same distinctive legal name. A supplier-identity record closing the ledger match is among the next records to pull.
Primary sources: City vendor payment ledger (published dataset); CA SOS entity record and filing history; CA AG Registry of Charitable Trusts. In the paid deliverable every material fact carries a document citation and a retrieval date.
Reproducible query: Aggregate the published ledger by vendor and fiscal year; compare against the SOS status history. Voucher-level dates require a records request and are listed under next records.
Pattern: Suspended / revoked while paid (observed in 8+ cases)

Alternative explanations tested

Payments recorded in the fiscal year may predate the April suspension; the published ledger carries no dates, so the split is unresolved.
Payments after the suspension date may cover services rendered earlier, tested by comparing service period to payment date once dated records arrive.
The organization was in a supervised windup (charity-registry dissolution waiver), and the city may have known and managed the wind-down.

Recommended next step

Request voucher-level payment dates and contract certifications from the Controller; refer status reconciliation to the funding departments. Frame: the overlap warrants review; it is not an assertion of wrongdoing by any individual.

Behavioral Health case, $989M DPH spending universe. Figures as recorded in primary filings; re-verifiable from the cited records.

How it works

Three moves from an organization to a review you can defend.

FIG 01

Resolve the organization

Match names, EINs, registry numbers, addresses, and jurisdictions so each record attaches to the identified entity rather than a similar name.

FIG 02

Reconcile the primary record

Compare the same facts across filings, audits, registries, court records, procurement history, and government payments.

FIG 03

Prioritize human review

Return sourced, evidence-graded items with uncertainty, alternative explanations, and the next record that could resolve the question.

See the evidence trail · a real case

Follow one organization across relationships, payments, and filings.

A real public-record investigation from its evidence file. Click any card to inspect the organization, relationship, or primary record behind it.

Fig 04 · entity graph: Conduit Foundationclick a card

Examined & clearedC LEADPrimary recordContradiction

Source:hand-curated evidence file for the Operator / Conduit Foundation matter (San Francisco), built from the framework-rerun survivors and the 2026 forensic review. The Operator's own books were examined and cleared; C LEAD items concern the conduit nonprofit and its disclosures. Party names withheld on this public page; full identities and the underlying records are available on email request, no call required. Nothing here asserts wrongdoing by any individual.

Case record to date

21 San Francisco cases and one BART investigation.

San Francisco is the main body of work. The BART investigation tested the method in a second jurisdiction.

Self-initiated public-record research, done on our own account. No client commissioned these investigations.

$11B+

Public spending reviewed

Public Works to Public Health: named department universes

Public-record investigations

21 San Francisco · 1 BART transit district

231

Organizations examined

42 affirmatively cleared on the record

88.5% of raw leads are discarded as noise before a professional reviewer sees them. How these figures are measured.

A CERTIFIED
Vendor 990 reports $0 government grants vs ~$28.2M on the city ledger in FY2021 alone
COVID emergency · San Francisco
A CERTIFIED
$74,247 paid to a vendor while FTB-suspended
Dream Keeper · San Francisco
A CERTIFIED
§1090 family conflict-of-interest matter (OIG 069-2022)
Construction mgmt · BART

What your team gets

Research your team can trace and reperform.

Four review checks for each promoted item

Source re-verificationCOMPLETED
Innocent-explanation testCOMPLETED
Opposing-counsel reviewCOMPLETED
Working-paper standardsCOMPLETED

88.5%of raw leads removed before professional review

Traceable

Every surviving item is challenged, linked to its primary records, and handed to a reviewer with uncertainty still visible.

$11B+primary-record review of complete recipient universes

Complete

Review the vendor or grantee population in scope across identity, standing, relationships, filings, payments, and governance.

Weeks of workassembled into one sourced memo

Fast

Reckton assembles public records into a sourced starting point while your team retains professional judgment.

1case record that carries forward

Reusable

Keep the organization, source set, comparison, review notes, and next step together for later funding, renewal, monitoring, or investigation.

Evidence before determination

Each review item includes its limits.

The six rules below keep sources, uncertainty, and reviewer responsibility visible.

Indicators are not determinations

A pattern or outlier is a reason to pull a document and review the context, never a conclusion by itself.

Every material fact traces to a source

Each figure, status, and relationship points to a filing, public record, or query a reviewer can reproduce.

Identity by hard identifiers

Organizations are matched on EIN, registry number, address history, and jurisdiction, never by a silent name merge.

Uncertainty remains visible

Missing records, weak matches, and credible alternative explanations stay attached to the review item.

Clearances are documented too

Of 231 entities examined, 42 were affirmatively cleared on the record. That list ships with the work.

Human review owns the decision

Reckton organizes evidence and prioritizes follow-up. Your team decides what to clear, escalate, or investigate.

Due diligence, defined

Questions a professional reviewer should ask.

Scope, source, product boundaries, and the practical path to a public-records pilot.

01What does Reckton investigate?

Reckton investigates organizations and entities. A review can include legal identity, standing, officers and directors, related organizations, government payments, nonprofit filings, charity registrations, audits, procurement history, court records, ownership records, and other authoritative public documents relevant to the decision.

02Where does the information come from?

The first pass is built from primary public records: government payment systems, IRS filings, state registries, charity records, corporate status histories, audits, procurement records, court records, and other official sources. Each material review item retains its source, date, jurisdiction, and locator.

03Does Reckton determine whether an organization committed wrongdoing?

No. Reckton identifies inconsistencies, relationships, missing records, and risk indicators that may require professional review. It does not make legal, fraud, corruption, or misconduct determinations. Your team evaluates the evidence, considers alternative explanations, and makes the decision.

04How is this different from a public-records database?

A database returns records. Reckton resolves the organization across records, compares the same facts across sources and time periods, records what is missing or conflicting, and packages review-worthy conditions with the evidence needed to understand them.

05Can Reckton support review after an award or decision?

Yes. The same case structure can support grant monitoring, contractor review, renewal diligence, and follow-up investigations. Reckton preserves the prior source set and review history so later reviews start from documented work. Talk to us about the cadence and source coverage required for your workflow.

06Does Reckton require private or sensitive data?

No private data is required to begin. A pilot can start with public records only. Where an authorized team later supplies internal records, they are identified as agency-provided evidence, handled separately from public sources, and governed by the engagement terms.

07Who is Reckton designed for?

Reckton is designed for professional teams making high-consequence decisions about organizations, including government oversight, audit, procurement, grant monitoring, compliance, legal, investigative, and forensic accounting teams.

08Can we begin with a public-records pilot?

Yes. A typical starting point is one defined decision workflow, department, recipient population, or organization set. The briefing identifies the public-record boundary, review standard, expected deliverables, and the questions the pilot should answer.

09Can we place the memo in our workpapers?

Not automatically. The memo is structured so your team can evaluate and reperform the public-record work. Your firm determines whether the preparer, scope, procedures, sources, and evidence meet its requirements, and it retains responsibility for any use in an engagement.

10How do you address independence and conflicts?

Before accepting a scope, Reckton identifies known financial, business, or other relationships with the agency, program, and material parties in scope. The fee is fixed and outcome independent. Your firm evaluates those disclosures under its own policies and standards.

Start with one vendor

Check one vendor.

Name the vendor and jurisdiction. We check whether relevant public records are available, confirm the $500 fee and delivery date, then return the evidence sheet. Your team decides how to use it.

$500 for one named California vendor. The fee stays fixed whether or not the records show an unusual condition
Two business days once relevant public records are available
State and federal exclusion and debarment list checks
Every material fact includes a citation and retrieval date; the work starts with public records
A $2,500 Vendor Exception Memo covers one defined agency program
Each check uses the same review process used for a pilot across a vendor population
No engagement is disclosed and no material published without written consent
Oversight offices can use this form to request a finding briefing