Enterprise AI, from strategy to systems your business runs on.
Catalyst is our enterprise consultation. Five stages that begin inside your data, design and size the system your business actually needs, prove it against one agreed number on real traffic, build it into the systems you already run, and keep operating it long after go live. Every engagement is scoped to the company in front of us. None of it is a package.
- Diagnose
- Design
- Validate30 to 45 days
- Deliver
- OperateOngoing, no end date
The projects do not fail on the model. They fail on the data underneath it.
We are not the first people to notice this. Three findings from outside this company say the same thing, and all three are about the stage that happens before anyone writes a prompt.
of AI projects that are not supported by AI ready data will be abandoned, through 2026.
of generative AI projects were forecast to be abandoned after proof of concept, on poor data quality, weak risk controls, rising cost or unclear business value.
of enterprise generative AI pilots studied produced no measurable return on the profit and loss statement.
Four rooms we keep walking into. One of them is probably yours.
A proposal you cannot judge
Someone has put an AI project in front of you and nothing in the document tells you whether the design is sound, what drives the effort, or what happens to it in year two.
Five systems, no source of truth
The same customer exists in the CRM, the ERP, the booking sheet and two WhatsApp accounts, spelled four different ways, and nobody owns the answer.
A pilot that died in production
It worked beautifully in the demo, then met real customers, real Arabic, real edge cases and real load, and quietly stopped being used.
A board that wants an answer
You have been asked what AI means for this company, and what you owe them is a position with numbers behind it, not a vendor deck with a robot on the cover.
Catalyst. Five stages, and a system at the end of them.
Catalyst is the engagement, not a framework we show you and then set aside. Each stage ends in something you can hold: a decision, a measured number, or a working system. If a stage cannot produce one, it has not finished.
- Diagnose
Discover and Assess
We start in your data, not in a demo. Departments at every level, the systems they actually use, and an honest read on whether any of it is ready.
- Design
Envision and Size
Opportunities scored twice, on value and on feasibility, then the system designed and sized against what discovery actually found.
- Validate
Prove
A 30 to 45 day proof against one metric agreed in advance, running on real traffic rather than on a rehearsed demo.
- Deliver
Build and Integrate
The full system, wired into what you already run rather than standing beside it as a second place your team has to remember to check.
- Operate
Operate and Optimize
The stage with no end date. Language drifts, prices change, customers ask new things, and quality falls slowly enough that nobody inside notices.
What each stage does, what you do, and what you end up with.
- 01Diagnose
Discover and Assess
What happens
We start with your data: where it lives, how clean it is, who owns it, and what it can actually answer today. Then we sit with your departments at every level, not the leadership alone, because the person answering the phone knows things the org chart does not.
What you do
Give us access to the systems you already run, and time with the teams that touch customers. No preparation and no polished process maps. We would rather see it as it is.
What you end up with
A readiness position
What your data can support today, what has to be fixed before anything is built on it, and which processes are worth automating first.
Most companies are not ready when we arrive. That is normal, it is fixable, and hearing it in week one is easier than discovering it in month four.
- 02Design
Envision and Size
What happens
We design the system against what discovery found rather than what was hoped for, then size it in three tiers: small, medium or large. The tier describes how much system you are putting into production, what sits inside it and what deliberately does not.
What you do
Pick the tier, and pick the one number the proof will be judged on. The second choice matters more than the first.
What you end up with
A system design, sized
The system in plain language, the tier it sits in, the integrations it depends on, and the order the pieces are built in.
- 03Validate
Prove
How long30 to 45 days
What happens
One number, not a scorecard. We build the smallest version of the system that could move it, and we run it on real traffic, in real Arabic and English, with real customers.
What you do
Agree the number before we start, and hold us to it at the end.
What you end up with
The number, measured
The agreed metric before and after, on live data, with a written recommendation to continue, adjust or stop.
- 04Deliver
Build and Integrate
What happens
We build the full system and connect it to what you already run: the booking system, the CRM, the sheet the office actually works from. Integration is most of the work. An agent that cannot read your calendar is a demo, not a system.
What you do
Name the people who own each system we have to connect to, and put a small group from your team on it first. Break it before your customers do.
What you end up with
A live system and a written handover
Connected and running, with the handover written down: what it does, what it will not do, who to call, and how to switch it off.
- 05Operate
Operate and Optimize
How longOngoing, no end date
What happens
We run it. Monitoring on the system itself, changes as your business changes, and improvement drawn from the conversations that actually happen. Customers ask things nobody wrote down, and those questions become answers the system knows.
What you do
Tell us what changed on your side: new services, new staff, a new branch, a new rule. The rest is ours.
What you end up with
A running service and a review
A log of what changed and why, and a review where we bring the numbers instead of waiting to be asked for them.
This is the stage where most AI projects quietly die, and it is the one this company is built around.
The deliverable is a system that runs your business.
The engagement is judged on what is running at the end of it, not on what was written during it. Assessments, data maps and plans exist because a system built without them is the one that gets switched off in month four. They are the route. This is the destination.
A system that does the work
The agent, the dashboard, the pipeline or the model that carries a real process end to end, inside your business rather than beside it.
One source of truth underneath it
The data map is how we get there. What you end up with is a system that reads one answer instead of four, and a rule for which system wins when they disagree.
The right thing built first
Every opportunity scored on value and on feasibility. The two rankings are rarely the same list, and the order you build in decides whether the second project ever happens.
A system sized to your company
Small, medium or large, with what sits inside the tier and what deliberately sits outside it named before anyone starts building.
Proof on real traffic
One metric, one time box, one stop condition, written before anyone starts so nobody gets to move the target afterwards.
A design your regulator can look at
Where the data sits, who can see it, what would leave the country, and which parts your regulator will want in front of them before any of it runs.
Every document produced along the way is yours to keep, written so another supplier could act on it. If the honest answer is that you should not build anything this year, that is what the assessment will say, and you still keep the map.
One method, every sector we are asked into.
- Banking and financial services
- Insurance
- Healthcare and clinics
- Real estate
- Contracting and construction
- Hospitality and hotels
- Retail and e-commerce
- Food service
- Logistics and transport
- Energy and industrials
- Government and public sector
- Education and training
- Automotive
- Facility and home services
- Professional services
- Telecommunications
Three things a vendor will not say to you.
Your data is not ready
Demoing an agent is easy. Telling a client their data cannot support one yet is not, and it is the sentence that saves the year.
You asked for the wrong thing
The project on the agenda is often worth less than the one nobody has written down yet. We will show you both, ranked, and let you choose.
This does not need AI
Sometimes the honest answer is a rule, a form and a report. We will say so, and we would rather lose the build than put a model where one is not needed.
The people in the room are the people who build it.
A consultancy that sells with its senior people and delivers with its junior ones is a familiar shape. We assemble the team around what the engagement actually needs, so the mix changes from one client to the next. What does not change is that the people in the discovery sessions are the people who build the thing afterwards.
Fixed scope, and an end date.
- Shape of the engagement
- Scoped to the company in front of us. There is no standard package, because there is no standard company, and the scope is agreed in writing before anything starts.
- Length
- Custom to the engagement, agreed up front and written into the scope. There is no standard length published, because there is no standard engagement.
- Who is in the room
- Departments at every level, not only the executive sponsor. The people who do the work know where it actually breaks.
- What we do not do
- Discovery workshops that end in a slide deck. Diagnosis is the work, and it ends in a written position you can act on with us or without us.
The procurement questions, answered before you have to ask them.
Do we have to build with you afterwards?
No. Every document produced during the consultation is yours, written so another supplier could act on it. We would rather you take a clear assessment to a competitor than take a vague one from us into a build.
Who owns what you build?
You own the system, the data and the business logic. We build on open infrastructure, so there is no proprietary layer you would have to buy your way out of later.
Can the models run inside our own environment?
Yes, and we will design it that way where your regulator requires it. We will also tell you what it takes to run and to keep current before you commit, because that is usually the deciding factor.
Do you have certifications like SOC 2 or ISO 27001?
Not yet. We are in the process of certification, and until it is in hand we are not going to imply otherwise. What we can do meanwhile is design to the controls your own auditors and your regulator require, and document the design so your compliance team can review it directly.
How large a company do you work with?
The method scales down more comfortably than it scales up. If you are a single branch business, a consultation is probably more process than you need, and we will say so on the first call rather than put you through one.
What if the assessment says we are not ready?
Then it says so, in writing, with the specific gaps named and an order to fix them in. That outcome is a success for you and it costs us a build, which is precisely why the diagnosis is worth running before the build and not during it.
Start with the diagnosis. Decide about the build afterwards.
One call is enough for us to tell you whether a consultation is the right next step, or whether you should take one of the products straight into production and get on with it.