Where AI can help
AI creates value where work is repetitive, where knowledge is hard to find, or where customers wait. These are the areas I look at first.
- Repetitive operations and manual processes
- Customer support and response times
- Internal knowledge and documentation
- Sales and marketing workflows
- Content production and research
- Business intelligence and personalization
AI consulting services
AI Strategy & Roadmap
Identify opportunities, priorities, risks and the sequence in which they should be implemented.
AI Assistants & Agents
Design AI assistants and agent-based workflows around real business processes rather than demos. See Chatbot and assistant development for how these are scoped and built.
Workflow Automation
Connect AI to the systems a business already uses and remove repetitive manual work. See AI automation for how a process is judged worth automating.
LLM Integration
Plan and integrate LLM-based functionality into existing products and workflows.
AI Tool Selection
Evaluate build-versus-buy decisions and select tools based on business requirements, not hype.
AI Adoption
Help teams introduce AI responsibly and practically, with training where it is needed.
How I work
- Understand the business problem
- Audit the current workflow
- Identify AI opportunities
- Prioritize by impact and feasibility
- Design the solution
- Implement or oversee implementation
- Measure and improve
When to bring in an AI consultant
Most businesses do not need an AI strategy. They need a decision about one or two specific things. These are the situations where an outside view usually earns its cost.
- You are being pitched AI tools and cannot tell which claims are real
- A department wants to automate something and nobody can judge whether it will work
- You have data you suspect is valuable and no idea how to use it
- A competitor has announced something and you need to know whether it matters
- An AI pilot ran, impressed everyone, and never reached production
- You need to know what AI cannot do before committing budget to what it can
Where the value usually is
Almost every conversation starts with a chatbot, and a chatbot is usually the least valuable thing you can build. The returns tend to sit in operations — document-heavy work, forecasting, scheduling, anomaly detection and knowledge retrieval — where the data already exists and the saving is measurable. I set out that argument in full in transforming operations with AI.
How an engagement works
- Diagnostic. What the business is trying to achieve, what systems exist, and what data is actually being captured today.
- Shortlist. Two or three candidate use cases, each tested against volume, repeatability, data trail and tolerance for error.
- Decision. What to build, what to buy, what to leave alone, and in what order — with the reasoning written down.
- Oversight. Review of whoever builds it, so the result matches the decision rather than the demo.
Consulting or buying a product?
| An AI product vendor | AI consulting | |
|---|---|---|
| Answers | Whether their tool fits | Whether you should build this at all |
| Incentive | Selling the licence | The decision being right |
| Best when | You already know what you need | The requirement is still a guess |
| Risk | You buy a solution to the wrong problem | Advice without delivery capacity behind it |
Both have their place. The failure mode I see most often in Kuwait is a business asking a vendor whether it needs the vendor.
What I will tell you not to do
If a use case will not pay for itself, I will say so. If your data cannot support what you want, I will say that instead of selling a data project to make it possible. If an off-the-shelf tool solves it for a monthly fee, that is the recommendation — it is a shorter engagement and it is the honest answer.
Frequently asked questions
What does an AI consultant actually do?
Works out which parts of a business AI can genuinely improve, decides what to build or buy and in what order, and oversees delivery so the result matches the decision. In practice much of the work is subtraction — ruling out the ideas that will not pay for themselves before anyone spends money on them.
Is our business too small for AI?
Size matters less than repetition. If something happens hundreds of times a month, leaves a data trail and follows rules a person could write down, it is a candidate — whether you have twelve staff or twelve hundred. What rules businesses out is usually absent data, not headcount.
Do we need to hire data scientists first?
Almost never for a first project. Models are rented, and the real work is integration, data quality and process design. Hiring specialists before you have a use case with a data trail is a common and expensive way to begin.
How well does AI work in Arabic?
Well enough to be useful, and still behind English. Modern Standard Arabic is handled reasonably, Gulf dialect less so, and text that mixes Arabic with English mid-sentence is hardest. Anything customer-facing in Arabic should be evaluated on your own material rather than on a vendor demo.
How do engagements start and what do they cost?
With a short diagnostic rather than a long programme — cheaper for you, and it means the first recommendation is based on what your systems actually do. Scope and cost are agreed before anything begins, and an audit or tool evaluation is a defined piece of work with an end date.