We build AI tools that tell you what they cannot do
Lulu Intelligence is a small practice based in Dubai. We work with teams in the region who want AI tools built carefully — with documented limits and human oversight at every decision point.
← Back to HomeHow Lulu Intelligence came to be
Lulu Intelligence started in 2021, after two of our founding team members spent several years working inside enterprise technology teams that were trying to adopt AI tools without a clear understanding of where those tools would fail. The projects were not disasters — but they kept producing the same pattern: tools that worked well in demos and poorly in practice, with no one who could explain the gap.
We started this practice to fill that gap. Our work is not about building the most sophisticated system — it is about building one that fits the actual work, that its users understand, and that does not create new problems while solving old ones.
We are based in Dubai because the UAE market is active, because the team is here, and because clients in the region often have fewer options than they would in London or Singapore when they need this kind of careful, scoped AI work.
What we are trying to do
People behind the work
Eight years designing machine learning pipelines for industrial inspection and logistics, most recently at a supply chain technology company based in Abu Dhabi.
Previously led NLP engineering for a legal technology firm in Dubai. Focused on document retrieval, citation systems, and the limits of language model outputs in professional contexts.
Leads architecture review engagements. Background in software engineering and data infrastructure, with particular interest in AI systems that operate in regulated or safety-adjacent environments.
How we run our engagements
Client data isolation
Documents and image datasets provided by clients are used only for the agreed project. They are not shared across clients and are not used to train shared models.
Defined scope in writing
Every engagement starts with a written scope document. Both parties agree on what the engagement covers before any work begins. Changes are discussed before being done.
Honest performance metrics
We evaluate model performance using test sets held out from training. We report precision, recall, and failure modes — not just headline accuracy numbers.
Limit documentation
Every delivered system includes a written note on what it handles reliably, what it handles partially, and what it should not be asked to do under any circumstances.
Human-in-the-loop design
We identify the decision point that matters most in each workflow and ensure a human operator can see, question, and override the system's output at that point.
Post-delivery reviews
For conversational interface projects, we conduct quarterly review sessions with the client team in the first year to assess whether the system is performing as expected in real use.
Applied AI work in the UAE context
The AI consulting market in Dubai and the broader UAE tends to attract large vendors with pre-packaged solutions and small advisory firms with more enthusiasm than technical depth. Lulu Intelligence sits between those two — small enough to work closely with one client at a time, technical enough to build the systems we recommend.
Our three service areas — computer vision tooling, internal conversational interfaces, and architecture review — reflect the most common requests we receive from teams in the region. Industrial and logistics operations need inspection tools that fit their workflow without replacing the operators who understand it. Corporate teams need internal knowledge interfaces that cite their sources. And technical leaders planning AI projects need a candid outside perspective before they commit significant engineering time.
We work in English and can work with Arabic-language document sets for retrieval and interface projects. Our architecture review sessions can be conducted in Arabic or English depending on the client's preference.
We do not subcontract or offshore any part of the technical work. Every engagement is handled by the team you speak to at the start.
Describe what you are working on
We will read your message, ask a few questions if needed, and let you know whether this is something we can help with and how.
Get in Touch