WHT-01 · FlowCount WHT

WHT Health Check

Find the withholding tax you missed before ZATCA does.

فحص ضريبة الاستقطاع: اكتشف المخاطر قبل الهيئة

Pilot availableFor: Head of Tax · CFO · Tax partner

1.50M

exposure found in a 37-line sample review

Basis: illustrative sample engagement (fictional data)

The problem

Payments to non-residents are coded by GL account, not by substance. Treaty relief gets applied without a valid certificate, and fines accrue at 1% of the unpaid tax every 30 days. The payer is personally liable.

“Your ERP codes WHT by GL account. ZATCA assesses it by substance.”

What FlowCount does

Recompute WHT on three to five years of non-resident payments; classify each payment by substance; test treaty, PE, timeliness and fines; rank the exposure.

AI and rules used
LLM classification with evidence; deterministic rules engine
Builds on
Built end to end in sample form

How it works

Who does each step

  1. 01Rules

    Prove the population is whole

    Tie every non-resident payment back to the trial balance before testing a single rate.

  2. 02AI

    Classify by substance

    Read contract, invoice and PO text; propose a WHT category with the quoted words it relied on.

  3. 03Rules

    Recompute, date-aware

    Effective-dated rates, treaty and PE tests, gross-up, fines to your as-of date.

  4. 04Reviewer

    Conclude and sign

    A named professional decides every judgement call; findings in 5C form.

What you send

  • AP payments to non-residents (30-field template)
  • Trial balance
  • Form Q history
  • Contracts and TRCs for top spend

What you receive

  • Exposure ranked by risk, tax + fines
  • Workpaper file with live formulas
  • Findings report and remediation plan
  • Refund schedule, reported separately

Controls built in

  • Completeness tie-out to TB (G2)
  • AI auto-accept only at confidence ≥ 0.85 and agreeing with GL
  • Workbook ↔ engine tie-out to the cent

KPIs we report

  • Cycle time G1 → G6 ≤ 15 working days (mid-size)
  • Tie-out failures at G4: 0
  • Findings later raised by ZATCA that we missed: 0

Sample input → expected output

What the result looks like

Basis: illustrative sample engagement (fictional data)

Headline from the sample engagement (37 lines, 28.1M, Jul 2023 to Jun 2026)
MeasureAmount
Under-withheld tax1,070,815.13
Delay fines accrued to 30-Sep-2026131,080.71
Potential WHT on a management fee paid outside AP (found only by the TB tie-out)300,000.00
Total exposure1,501,895.84
Over-withheld: refund candidates (reported separately, never netted)104,068.99
Duplicate payment found along the way337,500.00

Portfolio scorecard

Where it sits on the roadmap

Value

4.55 / 5

Feasibility

4.55 / 5

Priority index

6.90 · rank 9 of 27

Quadrant

Quick win

Delivery

Wave 1 (0-3 m)

T-shirt size M: 4 to 8 weeks with 2 engineers + 1 tax SME + 0.25 QA, indicative build 120k to 275k. New data source or document type + rules + review-UI changes. How we score and sequence →

Find the withholding tax you missed before ZATCA does.

A pilot runs on your own data, under the full G0 to G7 gate model, with a partner signature on every conclusion.