Product
One system for the work around every deal.
Zolia brings together what a brokerage runs on: leads, deal files, commissions, compliance and reports, with an AI team that works within clear rules. In English and Arabic.
AI team
An AI team for every brokerage, with guardrails
Every brokerage gets its own AI organisation: a management AI for the owner, an organiser with a fixed workflow, ten specialist agents and a personal assistant for every broker. Each agent has its own identity, a responsible person, a budget in AI units and an emergency stop.
- Permissions before the model: every data access and tool call is checked against the agent's permissions. Agents do not inherit the rights of the person who approved them.
- Autonomy you set, per agent and type of action: from shadow mode and suggestions to drafts for approval, acting with notice and acting with spot checks.
- Fixed caps: money actions go no further than drafts for approval, AML decisions no further than suggestions.
- A control agent checks samples of the others' work, and an agent that makes mistakes is moved down a level automatically.
- Signatures in DLD systems, payouts, decisions of the MLRO and HR decisions always stay with people.
Leads
Every enquiry answered, in the client's language
Leads from your website forms and your office WhatsApp number arrive in one place, including the ad reference of click-to-WhatsApp leads. They are routed fairly, and a register shows who owns each lead.
- The AI drafts a first response in the language of the enquiry. A person approves it before it is sent.
- Personal data is pseudonymised before any AI model sees it.
- Response times are measured, and brokers get push notifications for new leads.
- Every draft is checked for topics, claims and data leaks. In 226 test replies to enquiries, the AI invented no facts.
Deal files
Complete deal files, with KYC built in
Every deal has a file with checklists, document status and a KYC case for each buyer and seller. A deal cannot reach transfer without completed KYC.
- Uploaded documents are classified and their fields extracted, with a confidence check per field. Anything uncertain is held back for a person. In tests, 99.2% of 383 fields were read correctly across English, Arabic and bilingual documents.
- A source-of-funds questionnaire for every client, cross-checked by AI. Enhanced due diligence is approved by management.
- Introductions of clients are documented with a client confirmation in English and Arabic.
Commission ledger
Every commission, from deal to payout
The commission ledger follows every dirham: commission plans, receivables, payments, the split from gross to net, and clawbacks with a broker debt account.
- A cancelled deal stops open receivables and proposes a clawback for commission already paid.
- Off-plan: buyer payment plans, commission tranches by milestone, and an invoice draft for every tranche that falls due.
- A clawback radar for every deal, payout runs, and a promised payout time counted from receipt of funds.
- Brokers see for every deal what is expected, due, received and paid out. Payouts are always released by a person.
Compliance and AML
Compliance in every deal, decisions with your MLRO
Each brokerage names a compliance officer and a deputy. A stop button blocks every money step of a deal, and a compliance queue keeps deadlines and samples in view.
- One gate checks every outgoing client contact, by people or AI, against rules stored as data: no marketing without opt-in, never to bought lists.
- KYC, sanctions and PEP screening with daily re-checks, and drafts for goAML reports. Your MLRO decides.
- Flags for real estate activity reports based on payment method and amount, for example cash or virtual assets.
- AML by design: only the compliance agent can read the AML area, and it has no client channel. Other agents see only “compliance hold”, which prevents tipping-off.
- A consent register governs all client communication.
Reports
Reports for management, from fixed metrics
The management AI prepares a morning briefing and a monthly report with commentary for the owner. Every figure comes from a defined metrics catalogue, for example gross commission income, receivables by age and clawback risk.
- Every figure links to its definition, its source and its data date.
- The AI explains figures. It does not produce figures of its own.
E-invoicing
E-invoices following PINT AE
Invoices in Zolia follow the data model of the UAE PINT AE specification. The invoice XML is validated against the official validation rules and examples.
- An invoice draft for every commission tranche that falls due.
- Designed for transmission through an accredited service provider, ahead of the UAE e-invoicing mandate.
English and Arabic
Fully bilingual, right to left included
The interface works in English and Arabic, with a complete right-to-left layout. The AI answers clients in the language they wrote in.
- Documents are read in English, Arabic and bilingual form.
- Defences against prompt injection are tested in English and Gulf Arabic.
Quality, measured
Measured before release, not just promised
Every AI function has its own evaluation suite, and the whole system runs through automated tests on every change.
- 3,524automated tests on every change
- 99.2%of 383 document fields read correctly, in English, Arabic and bilingual documents
- 26prompt-injection attack patterns tested, in English and Gulf Arabic
- 0invented facts in 226 test replies to enquiries
As of 2 October 2026.
In development
Market data with sources
Next we are building market-data features for brokers' daily work:
- Price ranges with their sources, labelled as indicative, not a valuation, and shown only with at least five comparables.
- Rental index checks before renewal offers.
- Flags for listings of units that were already transferred.
- Market reports with traceable numbers.
See what it does for your brokerage
Start with the free 14-day Lead-Check, or tell us that you are interested in the pilot.