ACC-04 · FlowCount Assure

GL tax risk scanner

Score every line in the ledger for tax risk, not just the top 25.

فحص دفتر الأستاذ العام لمخاطر الضرائب، سطراً بسطر

Design partners wantedFor: CFO · Head of Tax · Internal audit

100%

of GL lines scored; the top 25 is only the default view

Basis: property of the method

The problem

Tax risk sits in the GL long before it reaches a return: entertainment and vehicle VAT claimed, cloud and software paid abroad without WHT, provisions and shareholder balances that move the zakat base. Nobody reads the whole ledger, so the review starts from a sample.

“The fastest health check is the one that runs on data you already have.”

What FlowCount does

Scan the full-period GL against VAT, WHT and zakat/CIT risk rules; rank every line by amount, rule severity, recurrence and confidence; show the top 25 by default while keeping every exception.

AI and rules used
Rules; LLM description classification; risk scoring
Builds on
FlowCount data contract

How it works

Who does each step

  1. 01Rules

    Score the population

    Every line tested against VAT, WHT and zakat/CIT rules. Nothing is capped.

  2. 02AI

    Read descriptions

    Classify free-text descriptions where the account alone is ambiguous, with evidence quotes.

  3. 03Rules

    Rank

    Amount, rule severity, recurrence and confidence. The top 25 are shown; all are retained.

  4. 04Reviewer

    Validate

    Confirm the top risks; false positives are marked and remembered for this client.

What you send

  • Full-period GL
  • Chart of accounts
  • Vendor and customer master
  • Tax codes
  • Invoices (optional)

What you receive

  • Top tax risks: transaction, amount, tax, risk, reason, evidence, recommended action
  • The full exception population
  • Recommended next review (VAT, WHT or zakat)

Controls built in

  • The full population is scored; the top 25 is a view, not a cap
  • False-positive marks change client rules, never statutory logic

KPIs we report

  • Share of flagged value confirmed by the reviewer
  • False-positive rate by rule

Sample input → expected output

What the result looks like

Basis: property of the method

Risk areas scanned
TaxWhat the scanner looks for
VATBlocked input VAT, entertainment, employee expenses, vehicles, related parties, imports, reverse charge, zero-rated revenue
WHTOverseas suppliers, management fees, technical services, royalties, software and cloud services
Zakat & CITProvisions, accruals, related parties, financing, fixed assets, investments, shareholder balances
Expected output: top risks (illustrative)
TransactionAmountTaxRiskReasonRecommended action
JV-2025-09311,500,000.00WHTHighManagement fee to a foreign parent settled by intercompany netting; no WHT bookedRun the WHT Health Check on related-party fees
AP-77120128,000.00VATHighCloud services from a non-resident booked to a domestic VAT codeSelf-assess the reverse charge; correct the tax code
AP-8041246,000.00VATMediumStaff entertainment with input VAT claimedReverse the claim unless a firm position allows it
GL 28102,400,000.00ZakatMediumProvision up 60% with no matching utilisationConfirm the provision before the zakat base is set

Portfolio scorecard

Where it sits on the roadmap

Value

3.90 / 5

Feasibility

4.50 / 5

Priority index

8.78 · rank 5 of 27

Quadrant

Quick win

Delivery

Wave 1 (0-3 m)

T-shirt size S: 2 to 4 weeks with 1-2 engineers + 0.5 tax SME, indicative build 50k to 120k. New rule set on the existing data model, one new input. How we score and sequence →

Score every line in the ledger for tax risk, not just the top 25.

Design partners shape the rulebook and review console, and get first access on pilot terms.