For DIFC & ADGM accounting firms
Draft client emails, summarize engagement correspondence, and pressure-test workpapers with GPT, Claude and Gemini in one workspace built for confidential practice. Client names, organisations, addresses, phone numbers and emails are masked before the model sees the prompt — in English and Arabic — and every query lands in a tamper-evident audit trail. Live this afternoon; no IT project.
Email signup, two minutes, no card required. No procurement, no installation.
An auditor's prompt — as the model receives it
Your audit record, per query
The three objections we hear in every firm conversation — in the order they come.
"Does client-confidential text really never reach OpenAI or Anthropic?"
Identifiers — names, organisations, addresses, phone numbers, emails, in English and Arabic — are replaced with inert tokens before the call; the model answers the anonymized text and your private copy is restored in your workspace. It fails closed: if masking can't run, the request stops. Detection is probabilistic (like every vendor's) — which is why every prompt also sits behind an audit record.
"What do we show a client — or a regulator — if asked?"
Every query your people make is recorded — who, when, which model, what was masked — in a hash-chained audit trail where each entry locks the previous one. Tamper-evident by construction: you hand over evidence, not a policy deck.
"We're 30 people with one part-time IT guy. Who operates this?"
You do, from a browser. One email signup creates the firm's workspace; the partner invites staff from the admin console and controls which models each person can use. It runs on Cloudflare's regional edge serving the UAE — nothing to install, no servers to own.
Four stages, every message. This is the whole mechanism.
Identifiers in your prompt become inert tokens before anything is sent. English and Arabic covered; requests that can't be masked are stopped, not slipped through.
Any model in the catalog — GPT, Claude, Gemini and more — answers the anonymized prompt. Pick by cost or capability per task.
Optional council mode for judgment calls: several models answer, review each other anonymously, and a chairman synthesizes one considered answer.
Tokens return to real names for your eyes only, and the event is appended to your firm's audit chain.
Pay-as-you-go credits — $1 = 100 credits, no expiry. Practices with 10–25 active users typically land between $50 and $200 a month — busy season included.
$0 to start.
Make it the version where the client is anonymized, the model is chosen, and every prompt is on the record — before a client's diligence questionnaire asks what you allow.
Start free — $5 credits, no card