Readiness Sprint
Find out what to build. A ranked list of AI use cases for your business, each with its cost, the data it needs and a roadmap.
You receiveA Readiness Report signed off by a senior engineer
AI engineering · GCC, US & Europe
We build AI assistants and agents that make it to production.
AI assistants, predictive models, location intelligence and computer vision. From strategy to launch in weeks, deployed in your region, signed off by a senior engineer.
Run an agency? Sell AI under your brand- 13,212Arabic & English documents in one assistant
- 24 hsaved every week by back-office agents
- 37GCC websites benchmarked
Built on
- Azure
- AWS
- Google Cloud
- Databricks
- LangGraph
- PyTorch
- MLflow
Proof of capability
Built by us. Running now.
Two tools our team built the way we build for clients. Try them on your own company.
AI Studioyour website, as an AI assistant
Enter your website. Ask it anything and see the sources, then get an AI opportunity scan of your business.
Run it on my websiteConsultant Toolkit28 AI skills, in your brand
The methods we use on engagements, packaged for Claude, ChatGPT and Copilot: board decks, business cases, RFP responses and executive Arabic.
Explore the toolkitWhat we build
Four kinds of AI, one senior team.
Scroll through the four. The scenes are illustrative; the numbers come from our projects.
01 / 04
Assistants & agents
AI that answers customers and staff from your own documents, and agents that handle multi-step back-office work.
13,212documents in one Arabic & English assistant
02 / 04
Predictive models & forecasting
Demand, budget and supply-chain forecasts, churn prediction and what-if scenarios your finance team can test.
18%less customer churn from a weekly prediction pipeline
03 / 04
Location & movement intelligence
Turn location and visit data into footfall patterns, catchment areas and audience profiles.
1.8Blocation signals turned into audience profiles
04 / 04
Vision on cameras & images
Detect, count and flag events in camera feeds and images, processed on-site so footage stays private.
82%accuracy classifying clinical images in testing
One team for all four, signed off by a senior engineer.
All of it runs on data platforms we have built at scale: Spark and Databricks pipelines processing billions of records, and dashboards your team already uses.
Services
Three ways to start.
Not sure what AI should do for you? Start with a one-week Sprint. Already know? Go straight to a build.
AI Readiness ReportSample · redacted
1 · Ranked use cases
| Value | Feasibility | |
|---|---|---|
| Arabic contract Q&A | High | High |
| Invoice matching | High | Med |
| Customer email triage | Med | High |
2 · Architecture sketch
Your data→Retrieval→Model→Evals
3 · Costed roadmap
- Phase 1 · MVP4 wk
- Phase 2 · Production3 wk
- Runmonthly
Signed off by a senior engineerSIGNED OFF · 08
Build
A working AI assistant or agent on your own data, tested in Arabic and English, running in your cloud. Or we take your prototype to production.
You receiveA live system with documentation and test results
Run
We keep it running and improving, with a weekly quality and cost report. Or we act as your part-time Head of AI.
You receiveA weekly report and one accountable engineer
Fixed scope and fee, agreed in writing before we start. We reply within 24 hours.
Selected work
Problems our engineers have already solved.
Delivered by our engineers in previous consulting work. Client names stay private.
- 01Government · GCC
Arabic document assistant for a public-sector body
13,212documents searched, with the exact paragraph cited
- The problem
- Staff spent hours searching thousands of Arabic and English regulations, circulars and reports.
- What we built
- A retrieval assistant that answers in Arabic or English and cites the exact paragraph, running inside the client's own cloud region.
- The result
- Staff go straight to the source paragraph across 13,212 documents, and the data never leaves the client's region.
- 02Energy & retail · GCC
Operations analytics across a national network of fuel stations
- The problem
- Hundreds of sites, each with its own cameras and sales data, and no single view of operations.
- What we built
- Vision models that turn camera feeds into events such as queues, vehicle counts and safety issues, joined with sales data in one dashboard.
- The result
- One view of every site, with footage kept on-site, which made privacy approval far easier.
- 03Operations
AI agents for back-office workflows
24 hsaved every week
- The problem
- Repetitive multi-step work across email, documents and internal systems tied up skilled staff.
- What we built
- Agents that read requests, pull data from internal systems and draft the output, with a person approving every step that can't be undone.
- The result
- 24 hours a week of re-keying saved. Skilled staff review and approve instead, and every agent step is logged.
- 04Adtech
Audience profiles from location data
1.8Blocation signals processed
- The problem
- Advertisers knew little about who visited which places, so campaigns targeted broad, expensive segments.
- What we built
- Spark pipelines on Databricks that turn location-visit patterns into audience profiles, refreshed at scale.
- The result
- 1.8 billion location signals from 20 million devices became audience segments advertisers can target.
- 05Public sector · GCC
Budget forecasting for a government investment fund
- The problem
- Budget planning for a large investment portfolio relied on spreadsheets that could not test what-if scenarios.
- What we built
- A predictive model with built-in scenario analysis, so planners compare outcomes before committing funds.
- The result
- More accurate budget forecasts, with every scenario documented for review.
How we work
Seven AI agents do the work. The eighth signs it off.
Our engineers use AI tools to move faster at every stage. A senior engineer makes the decisions, reviews the work and signs off every deliverable before it reaches you.
- 01
Discover
A 30-minute call, then a written proposal with scope, timeline and fee within 48 hours.
- 02
Design
System design: data, retrieval, models, infrastructure and cost, agreed with your team.
- 03
Build & evaluate
We build in your cloud and test accuracy, speed and cost in English and Arabic, with a weekly status note.
- 04
Sign-off & handover
A senior engineer reviews every line and every evaluation result. Then a full handover, or we run it for you.
- 1 wkReadiness Sprint
- 48 hWritten proposal
- MultilingualEnglishالعربيةGulf dialect
- 8 / 8Deliverables signed off
Trust & data
Your data stays under your control.
GCC organisations: Arabic-first. US & Europe: production fast.
- Deployed in your regionYour cloud or ours, in-country where required.
- NDA and data agreement firstSigned before we see any of your data.
- Evaluated in Arabic and EnglishIncluding dialects, before anything goes live.
- Senior engineer sign-offOn every deliverable, written into every contract.
FAQ
Questions clients ask.
Who is responsible for quality?
Our senior engineers. Nothing is delivered until a senior engineer has reviewed it and signed it off, and that sign-off is part of every contract.
Where is our data stored and processed?
In the region you choose: your own cloud account or ours, in-country where your rules require it. We sign an NDA and a data processing agreement before we see any data, and your data goes into an AI tool only with your written consent.
Can you work with our existing team?
Yes. We often work alongside in-house engineers, or as a white-label partner for agencies that need senior AI delivery.
Can you work white-label for my agency?
Yes. We build under your brand, you own the client relationship, and we never contact your clients. See how the partnership works.
How is an engagement priced?
Every engagement has a fixed scope and a fixed fee, agreed in writing before we start. After a 30-minute call you receive a written proposal within 48 hours.
Tell us what you want AI to do for you.
A 30-minute call with a senior engineer, in English or Arabic. A written proposal follows within 48 hours.
a real conversation, not a pitch