Recover more. Prove every step.

ClearGrid is the collections platform for banks, BNPLs, telecoms and lenders in the UAE and Saudi Arabia. We decide the right next move on every account, work it in Arabic and English across AI voice, messaging and self-service, and leave a record your compliance team can reconstruct.

Built for collections, by collections. Built by the region, for the region.

$1.2B in loans under management · UAE & Saudi Arabia · as at Q1 2026

Scroll
ClearGrid Command
Command modules
Command reporting: portfolio stats, settlements and the deal-stage funnel per lender Custom views, per lender The recovery funnel, live
01 · Reporting

The whole book, before you touch anything.

Recovered, committed, still owed — per lender or across all of them. Every number here is a filter, so the question becomes the queue.

Explore the module
Borrower scoring: call pickup score, RPC and PTP probability and the best contact window per borrower Top window 18:00–20:00 GST Pickup, RPC and PTP per row
02 · Borrower data

Pickup, RPC and PTP, scored per account.

Model probabilities and the best contact window for every borrower, refreshed daily, filterable and exportable.

Explore the module
Unified inbox: every inbound conversation in one queue, linked to its account with the assigned agent and stage One queue, not three tools Each thread attached to its account
03 · Inbound support

When they come to you, nothing is lost.

Every inbound route — inbox, tickets, disputes, calls — lands attached to its account, in a queue somebody owns.

Explore the module
Call history: manual, campaign, inbound and AI calls with outcomes, dispositions and recordings AI → human transfer, mid-call Manual and AI, filterable side by side
04 · Dialer

Every call, one history.

Manual, campaign, inbound and AI calls in the same log on the same definitions — the only way comparing them means anything.

Explore the module
Journey builder: entry trigger, contact-window condition, AI voice step, salary-cycle wait and a broken-promise fallback Journey GPT drafts the canvas Policy-checked before every send Broken promise → auto re-engage
05 · Journey builder

Eight block families. Or just describe it.

Visual journeys with draft→publish governance, and Journey GPT drafts the canvas from plain language.

Explore the module
Agent performance: promises secured, right-party rate, settled deals, the reached and not-reached outcome mix and per-agent activity Flagged calls awaiting review Per-agent, not per-team
06 · Agent workspace & analytics

The floor, and how it actually performed.

Promises secured, right-party rate, active time against idle time, per agent — measured the same way for people as for the AI.

Explore the module
ClearVoice live call: Gulf white Arabic detected, bilingual transcript, six policy gates passing and a promise to pay captured Dialect detected, not assumed Six gates, checked mid-call
06 · ClearVoice

Arabic on the left, English on the right, policy on both.

A live call with the dialect detected, every policy gate checked as it happens, and the transcript in both languages side by side.

Explore the module
Compliance and DNC audit: active suppressions, attempts blocked before dispatch, contact rules in force and the suppression register Blocked before dispatch, not after Ramadan schedule applied automatically
07 · Compliance & DNC audit

12,904 attempts stopped before dispatch.

Contact windows, frequency caps, Ramadan schedules and do-not-contact enforced by the engine — then exported as evidence.

Explore the module
QA Eval: quality score against target, 100% call coverage, four bands and failed checks named and counted by category AI 89 · human 87, same rubric Hallucination is a named, counted failure
08 · QA Eval

Every call scored. AI and human, one rubric.

Four bands instead of a pass mark, compliance failures counted separately, and AI-specific failure modes named — hallucination included.

Explore the module
Lender register: one sealed environment per portfolio, each with its own policy set, isolated data and account counts Zero cross-lender data events Its own database, per lender
09 · Lenders

Ten environments. Zero cross-lender events.

Each portfolio in its own sealed environment with its own database and policy set. The number that matters is the last one: zero.

Explore the module

That is Command, running a real book. Bring one portfolio slice and we will show you the same views on your own accounts.

Inside Command

Journeys · Journey GPT · AI Voice · Campaigns · Audiences · Scoring · Agent Workspace · Disputes · Unified Inbox · Unified QA · DNC Audit · Reports · Settlements · Skip Tracing · Deliverability · Multi-Lender

Live portfolio operations across the UAE and Saudi Arabia

+38%
increase in collections
2x
faster time to resolution
+60%
more borrower engagement than traditional agencies
4.8/5
borrower satisfaction
$1.2B
in loans under management
Your problem

Your agencies get paid to make calls. You need to get paid.

Most collections still runs on dial counts: pressure scripts, blind timing, no memory of the last conversation. Every month it burns customers you paid to acquire, and every quarter the provision line grows.

Problem 01

You are billed for activity, not recovery

Commission is charged on dials and letters. The incentive is volume, and volume is what you get — whether or not it collects.

Problem 02

The same treatment for every borrower

No scoring, no channel preference, no timing. The customer who would have paid on a message at 7pm gets the same call at 10am as the one who never will.

Problem 03

Nobody remembers the last conversation

A borrower who explained their situation last week explains it again. Every repeat is a customer you already paid to acquire, deciding not to come back.

Problem 04

You cannot evidence conduct

What was actually said on a call sits in an agent’s note field. When the regulator or your own conduct committee asks, there is nothing to show.

Problem 05

Adding callers is the only lever

Cost-to-collect rises with headcount because nothing in the operation learns. Next quarter starts from the same place as this one.

Problem 06

You are reporting on trust, not data

Allocations, payments and promises live in separate systems, so the recovery number is rebuilt by hand every month and nobody can audit it.

The ClearGrid solution

Change what gets measured.

ClearGrid works every account on outcome, not activity — scoring who to reach, choosing the channel and the hour, holding the context of every prior conversation, and evidencing all of it. It runs alongside your core systems, so nothing gets ripped out.

If any of those six describe your operation, the conversation is short. We model your current cost-to-collect against Command on a slice of your book, before you commit anything.

Three ways to work with us

However you collect today, we plug in.

One platform underneath. Start with one, add the rest when you are ready.

Customer experience

Better treatment produces better repayment.

Some borrowers cannot pay. Some will not. Some are confused, some are disputing the balance, and some simply forgot. We can tell the difference, and each one gets a different conversation. Arabic-native, on the channel that borrower actually answers, with human specialists on every hardship flag.

One engine · native in both languages

مرحباً أحمد، نذكّرك بخطة السداد المتفق عليها. يمكنك تأكيد الدفعة القادمة أو تعديل الخطة بما يناسبك.SMS · Arabic · 18:40 GST · inside contact window
Hi Ahmed, a reminder of your agreed instalment plan. You can confirm the next payment, or adjust the plan to fit your situation.SMS · English · policy-checked wording

Heard from ClearGrid about an account? Resolve it directly at cleargrid.ae.

How it works

A collections agency, supercharged by AI.

No system replacement and no big-bang cutover. We connect to your data, mirror the treatment policy you already have, and prove it on a slice before anything touches a live borrower.

01

Connect

Accounts flow in over SFTP or API. Your portfolio structure, buckets and reference IDs are preserved.

02

Prioritise

Every borrower is scored daily for how likely they are to pay and how likely a call is to connect. Effort follows likelihood, not queue order.

03

Engage

AI voice, SMS, email or a human specialist — in Arabic or English, only inside the calling hours and contact limits you set.

04

Resolve

Take the payment, set up the plan, or agree the settlement — within the offers you have authorised.

05

Report

Continuous reporting per account, and an audit trail your compliance team can reconstruct any conversation from.

Head to head

We ran our AI against human collectors on the same book.

Not a projection and not a vendor benchmark — a live test with a GCC lender, same portfolio, same period. Our AI voice and journeys recovered more than the human collection teams working alongside them, faster and at a fraction of the cost per account.

Trust

Built for compliance-first institutions.

All of it wrapped in a security and in-country hosting layer built for regulated GCC lenders.

Data stays in region

UAE portfolios hosted in-region. Saudi portfolios processed in-Kingdom, as committed in our DPA.

Independently audited

ISO 27001:2022 certified and SOC 2 attested. Certificates, scope and reports in the Trust Center.

Your policy, enforced by the engine

Contact windows, frequency caps and do-not-contact applied at decision time on every channel, not left to training.

Every conversation on the record

Calls recorded and quality-scored, with role-based access and an audit trail your compliance team can reconstruct.

Your CISO’s questions have documented answers. We will walk your security, compliance and conduct teams through all of it on the first call.

Outcomes

Recover more. Spend less. Keep your borrowers.

AED 3M
recovered at ≈ AED 26K execution cost in one live deployment
3% → 15%
answered-conversation conversion vs. the agency baseline
95%
of operations automated, with humans on judgement calls
Read the case studies Measured on live GCC portfolios against the prior agency baseline. Full methodology available on request.

Prove it on your own portfolio.

Start with a pilot slice, measure resolution against your current agencies, and see the numbers in weeks.

Book a consultation