KansaGrid
Book a demo
Audit-grade evidence for AI in finance

Prove every AI-driven financial action followed policy.

KansaGrid is the independent control-evidence layer for AI in finance. We connect every AI action — invoice, payment, journal, reconciliation, onboarding — to its source data, accounting policy, internal control, approval, exception and audit-grade evidence.

Book a demoHow it worksSOC 2 · ISAE 3402-aligned · audit-ready
evidence / TX-2026-0481 · replayable audit file
evidence sufficient
Replay
  1. 08:02:14Source

    Invoice received — vendor #VN-0149 · €48,230.00

  2. 08:02:15AI Action

    AP Agent matched PO-7781 · GL 6400 · WHT 5%

  3. 08:02:15Policy

    AP-POL-v3.2 · Three-way match · Cut-off Q1

  4. 08:02:16Controlexception

    CTRL-AP-04 Approval threshold > €25k

  5. 08:04:02Approval

    CFO override · justified · signed

  6. 08:04:03Posting

    ERP journal #JE-228144 booked

Audit assertions
Existencesatisfied
Completenesssatisfied
Accuracysatisfied
Cut-offsatisfied
Classificationsatisfied
Authorizationreview
Evidence sufficiencyv2026.04
94/1006 of 6 assertions covered
1 exception · 1 override · 0 gaps
hash · 7a4e…b21f · tamper-evidentExport pack ↗

Built with auditors, CFOs and regulators across regulated markets

Regulated bank
Tier-1 audit firm
Global fintech
Global asset manager
Payments platform
Wealth advisor
The problem

AI now touches financial reporting. Audit assurance hasn't caught up.

CFOs, internal audit and external auditors are being asked the same question by their boards and regulators: how do you know the AI did the right thing?

  • 01

    AI is moving into finance

    Invoices, payments, onboarding, revenue recognition, reconciliations and close are increasingly executed or accelerated by AI agents.

  • 02

    But audit didn't disappear

    Boards, auditors and regulators still need to prove every financial action followed approved accounting policy and internal control.

  • 03

    Logs are not evidence

    Chat transcripts, screenshots and prompt traces are not audit evidence. They don't map to assertions, controls or financial impact.

  • 04

    Generic AI governance falls short

    It governs model risk, not financial controls. It can't tell an auditor whether a journal entry is complete, accurate, classified and authorized.

The platform

The evidence layer for AI-driven finance.

KansaGrid encodes financial control methodology into software, so every AI action in finance is automatically traceable, replayable, controlled and audit-ready by design.

See how it works
Core IP

Financial Control Ontology

A machine-readable model that links every AI-finance action to accounting policy, internal control, audit assertion, approval, evidence and financial impact.

Methodology

Audit Assertion Mapping

Each AI action is mapped to the assertions auditors actually test: existence, completeness, accuracy, cut-off, classification, authorization and valuation.

Engine

Evidence Sufficiency Engine

Continuously evaluates whether sufficient and appropriate audit evidence exists for an AI-driven financial action — not just whether it was logged.

Trust

Tamper-evident Evidence Graph

Structured evidence with hashes, source references, identities, timestamps, policy versions and change logs. Stronger than PDFs, screenshots or chat logs.

Monitoring

Exception & Override Intelligence

Detects rare overrides, missing approvals, SoD conflicts, policy breaches, prompt/model changes and evidence gaps across workflows.

Architecture

Independent SDK & API

Runs above your AI agents, ERP, approval tools, SharePoint and finance systems. Vendor-neutral by design — we don't build the agent, we control it.

How it works

From AI action to audit-grade evidence, in five steps.

01

Connect

SDK and connectors plug into your AI agents, ERP, approval tools, SharePoint, Outlook/Teams, Salesforce and AI-agent frameworks — without rebuilding any workflow.

02

Capture

Every AI-finance action is captured as a structured event with source data, identity, prompt/model version, policy version, control reference and timestamps.

03

Control

Workflow-specific control packs evaluate the action against accounting policy, internal control and audit assertions — in real time.

04

Evidence

An immutable, tamper-evident evidence graph is generated per transaction — replayable end-to-end from source data to ERP posting.

05

Assure

Continuous monitoring surfaces exceptions, gaps and overrides. Export auditor- and regulator-ready evidence packs in one click.

Replayable Audit File

Replay any AI-driven transaction. End-to-end. On demand.

For any transaction, an auditor can replay the full chain: source data → AI output → policy and control checks → approval or override → ERP posting → evidence. No screenshots. No reconstruction. No ambiguity.

  • Deterministic, time-ordered event chain
  • Policy and control versions snapshotted at run-time
  • Cryptographic hashes link each step to source data
  • Exception, override and approval reasoning preserved
Source
AI · Policy · Control
Evidence
Vendor invoice #INV-9381
PO-7781 · GR-4419
Bank statement line #2298
AP Agent · v1.4.2
AP-POL v3.2 · cut-off Q1
CTRL-AP-04 · 3-way match
hash 7a4e…b21f
signed · CFO override
ERP JE-228144
1 exception · 1 override · 0 evidence gapsReplay ↗
Workflow control packs

Domain-specific control packs for every finance workflow.

Each pack ships with policies, controls, audit assertions, exception rules and evidence schemas — co-designed with RAs, internal audit leaders and CFOs.

Accounts Payable & Invoices

Three-way match, duplicate invoice detection, GL classification, withholding, VAT/IT-tax, cut-off and approval thresholds.

CompletenessAccuracyCut-offAuthorization

Payments & Treasury

Sanctions checks, beneficiary validation, payment authority, dual control, FX policy and supporting evidence per disbursement.

AuthorizationAccuracyExistence

Vendor & Customer Onboarding

KYC/KYB verification, beneficial ownership, sanctions/PEP, conflict of interest, contract policy and SoD checks.

Rights & ObligationsAuthorization

Revenue Recognition

Contract identification, performance obligation, transaction price, cut-off and accounting policy alignment to IFRS 15 / ASC 606.

CompletenessAccuracyCut-offClassification

Reconciliations

Bank, intercompany, sub-ledger and clearing reconciliations with explained variance, ageing and supporting evidence.

CompletenessExistenceAccuracy

Financial Close & Journals

Manual journal review, late adjustments, top-side entries, threshold review, materiality and disclosure support.

CompletenessClassificationCut-off
Continuous Control Monitoring

Periodic sampling is over. Welcome to always-on assurance.

Monitor every AI-driven workflow continuously. Surface missing evidence, failed controls, unusual overrides, SoD conflicts, prompt or model drift — before they become audit findings.

12
Workflows
47
Assertions
92
Avg evidence score
Live exceptionslast 24h

  • Payment authorized without dual control

    Treasury · Payments · Authorization

    open

  • Three-way match override without justification

    Accounts Payable · Accuracy

    open

  • Vendor onboarded with stale KYC evidence

    Vendor onboarding · Rights & Obligations

    open

  • Manual journal posted after cut-off window

    Financial close · Cut-off

    open

  • Prompt/model version changed mid-close

    AP Agent · Methodology drift

    open
Who it's for

One independent evidence layer. Four audiences that finally agree.

CFO & Controller

Adopt AI in finance — without losing control.

Roll out AI agents across AP, payments, close and reconciliations with continuous proof that approved policy and controls were followed.

↓ 40%
audit prep time
Internal Audit

Continuous assurance, not periodic sampling.

Test every transaction against every relevant control. Investigate exceptions instead of building working papers from screenshots.

100%
population coverage
External Auditor

Audit AI-driven workflows with confidence.

Standardized, methodology-aligned evidence per assertion. Replayable transactions. Tamper-evident artefacts. ISAE-3402-ready packs.

→ assert.
evidence per assertion
Audit Committee & Regulator

A trustworthy record of AI in financial reporting.

An independent, machine-readable record of every AI-finance action, control execution, exception and override — across all entities.

1 source
of AI-finance truth
Methodology

Encoded by auditors. Validated with regulators.

Our Financial Control Ontology is built with Registered Auditors, internal audit leaders, audit partners, advisory firms, CFOs and regulators. We don't advise on AI governance — we encode financial control methodology into software and continuously prove whether AI-driven finance workflows followed it.

  • Aligned with ISA 315, ISA 330, ISAE 3402 and SOC 2
  • Mapped to IFRS, US GAAP, IAS 1, IFRS 15 and IFRS 9
  • Methodology validation network of RAs and audit partners
  • Benchmark dataset across workflows, controls and exceptions
Audit assertion ↔ AI action map
AssertionExample AI actionEvidence
CompletenessAP agent posts all matched invoicesSource ↔ ERP delta proof
AccuracyGL classification & VAT calcPolicy ref + recomputation
Cut-offQ1 close booking decisionsTimestamp + period stamp
ClassificationCapex vs opex routingDecision trace + policy
AuthorizationPayment > €25k releasedApprover identity + signature
ExistenceVendor onboarding completedKYC artefacts + ownership
Where we start

The regulated-finance wedge.

We launch with regulated financial services — banks, fintechs, payments, wealth, insurance and asset managers — where the regulatory bar is high and AI adoption in finance is accelerating fast.

  • · Configurable, region-aware data residency and review
  • · Designed around prevailing financial-regulator expectations
  • · Co-built with Big-4 and mid-tier audit partners
Why now

  • AI hits financial reporting

    Boards and regulators are asking for proof, not promises.

  • Logs are not evidence

    Audit needs structured, replayable, methodology-aligned artefacts.

  • Independence matters

    The agent's own logs cannot be the audit evidence.

Integrations

Vendor-neutral by design.

We don't replace your ERP, your AI agents or your approval tools. We sit alongside them — capturing structured evidence and applying controls without forcing migrations or rebuilds.

ERP & Finance systems
SAPOracle FusionNetSuiteMicrosoft DynamicsWorkday
Approval & workflow
CoupaSAP ConcurAribaServiceNowDocuSign
Collaboration & evidence
SharePointOneDriveOutlookTeamsSlack
CRM & front-office
SalesforceHubSpotDynamics CRM
AI agent frameworks
OpenAIAnthropicAzure AILangChainLlamaIndex
Identity & access
OktaMicrosoft EntraAuth0
Vision

From periodic sampling to continuous, machine-readable assurance.

AI is moving into financial reporting. Audit, internal control and board accountability are not going away. As AI agents enter finance operations, companies need a system of record proving every AI-driven financial action followed approved controls, policies and audit methodology.

In an AI-run company, every financial action should be automatically traceable, replayable, controlled and audit-ready by design. Financial audit will move from periodic human sampling to continuous, machine-readable assurance.

We are building the control evidence infrastructure that makes that transition possible.

Get started

Make your AI-driven finance audit-ready.

Talk to us about a 6-week pilot on one finance workflow. We'll capture evidence, run the control packs and deliver an audit-grade evidence pack for your next close.

Book a demohello@kansagrid.com
Pilot in 6 weeks
  1. W1 · Scope workflow & control pack
  2. W2 · Connect AI agents, ERP & approvals
  3. W3–4 · Capture evidence on live operations
  4. W5 · Replay, exceptions & remediation
  5. W6 · Auditor-ready evidence pack