Product

Eight practice areas, twenty-two modules

One shell, eight practice areas, 22 modules you switch on as you grow — plus a separate client portal. Every area runs on the same trust model: AI drafts, rules compute, a person signs, the platform remembers.

All screenshots show the demo tenant (Nakhla Foods) — fictional data, captured from the 4-minute demo.

Bookkeeping

From documents to draft books

Bulk intake from Excel, PDF or XML — Arabic and English alike — with OCR confidence scores on every field.
The AI drafts a balanced journal, in purple, that does not post until a reviewer accepts; every field links back to the exact spot on the source document.
Years of paper backfile in monitored batches, with a cost-guard and a human QC queue for anything low-confidence.

Voucher OCRHistorical DigitizationEvidence LibraryDocument Sorting

Voucher OCR screen: an AI-suggested journal entry, labelled as a draft, with Accept, Modify and Reject buttons — Voucher OCR — the AI's journal draft posts only when a reviewer accepts. Demo tenant, fictional data.

Compliance & Reports

ZATCA and IFRS, computed — not guessed

The rules engine checks each invoice against ZATCA rules and returns exact error codes and fields — rules-checked before submission (sandbox during the pilot).
IFRS statements and notes are generated by versioned rules; unmapped accounts block the run, four reconciliations gate partner sign-off, and export is bilingual.
A deterministic 13-week cash forecast and AML / KYC with a rules-computed verdict round out the reporting stack.

ZATCA Mock ReviewIFRS Full PackFP&A / Cash FlowAML / KYC

ZATCA Mock Review screen: the rules engine returns exact error codes per invoice before submission — ZATCA checks — deterministic rules, exact error codes, caught before submission. Demo tenant, fictional data.

Audit & Risk

An audit trail you could hand to a regulator

Rules score every journal — one hundred percent of entries, not a sample — and show why each was flagged.
The AI explains the risk and suggests review questions; the product states plainly that high risk is not fraud.
Financial due diligence turns a data room into cited findings, isolated from the rest of the practice.

Journal RiskFraud & AnomalyFinancial DD

Journal Risk screen: journals ranked by rules-derived risk scores with a manual disposition gate — Journal Risk — scores are rules-derived, and every high item passes a human gate. Demo tenant, fictional data.

Tax

Zakat and VAT with the working shown

VAT, Zakat and WHT computed deterministically with citations — the model never computes the tax.
The Tax Copilot answers with the regulation, article and version cited — or refuses and routes to a specialist.
The Knowledge Copilot answers across regulations, client files and firm memos — permission-aware, never across tenants.

Tax CalculationTax CopilotKnowledge Copilot

Tax Copilot screen: draft answers with regulation citations, and a refusal when no source covers the question — Tax Copilot — cited answers, or a refusal routed to a specialist. Demo tenant, fictional data.

Operations

Run the practice, not the paperwork

The Practice Console is the partner's cockpit — open tasks, OCR batches, client blockers and one review-and-sign queue.
Realization, recovery and DSO at a glance, AR aging, and partner-gated billing proposals.
Engagement letters draft only after independence and pricing checks clear.

Practice ConsoleClients & TasksProposal & EngagementWIP & Billing

WIP & Billing screen: realization, recovery, DSO, AR aging and partner-gated billing proposals — WIP & Billing — the firm's own economics, partner-gated. Demo tenant, fictional data.

Governance

Control who sees and signs what

Deterministic independence and conflict checks return the verdict; the AI only explains — sending is blocked until conditions clear.
Ethical walls keep client teams separated by access control, not by memo.
Partner-only role management, with every action written to the append-only audit trail.

Independence & PDPLUsers & Audit

Independence & PDPL screen: a rules verdict with conditions, a UBO graph, and an AI memo labelled draft-only — Independence & PDPL — the verdict is rule-computed; the AI drafts the memo only. Demo tenant, fictional data.

AI Workspace

Drafting help that shows its sources

Ask for a task in plain language and get a review-ready draft — everything routes through the review gate.
Quick actions for OCR, ZATCA, IFRS and evidence, without leaving the workspace.
The Skill Hub switches the 22 modules on as you grow, each marked for its AI fit.

AI WorkbenchSkill Hub