Industrial AI practice  ·  Gulf  ·  Arabic and English Reply within one working day

Engagement model

four stages, and you can stop after any of them

01

Diagnose, then specify, then build

Each stage is a complete piece of work with its own deliverable. Each one ends in a decision, not an assumption.

You already bought a pilot once. It worked, and then it stopped, and nobody could say exactly why.

Building is no longer the hard part. Modern tooling has made it cheap and fast to build almost anything, which means the expensive decision is no longer how to build. It is what is worth building, in what order, and whether the organisation, the team and the data can absorb it. That judgment is what we sell.

Published estimates put the failure rate of enterprise AI projects somewhere between 70% and 95%. The figure is widely reported and we have not found a primary source we are willing to cite for it, so treat it as the direction of travel rather than a measurement. The cause is not in dispute: automating a process that was already broken, or building a capable system for a problem nobody quantified first.

Durations are typical, not promotional. No prices on this site: an engagement is priced against a defined outcome, in a written proposal, with three options to choose from. We do not bill hourly, and you always know the number before the work starts.
Stage The question it answers Typical duration
Stage 1 · DiagnoseReadiness & Opportunity Audit Is there a real, quantifiable problem here, which two or three opportunities are worth pursuing and in what order, and is the organisation ready? 2-4 weeks
Stage 2 · SpecifyStrategy & Roadmap How exactly do we do it, what does it cost, who does it, in what order, and what has to be true for it to work? 6-10 weeks
Stage 3 · BuildImplementation Does it work in production, and do your people actually use it? 3-12 months, phased
Stage 4 · OverseeGovernance & Advisory Are the systems still performing, what new risk has appeared, and is the roadmap still pointed at the right things? Ongoing, reviewed annually
02

Readiness & Opportunity Audit

Stage 1 · Diagnose · 2-4 weeks

A bounded diagnostic, not a survey. We interview the people who actually run the process, look at real data samples and real systems as they operate rather than as they are documented, establish a baseline where none exists, and score readiness across the dimensions that determine whether a deployment survives contact with a real working week.

What you receive

  • A ranked opportunity map, scored on impact, effort and data readiness
  • The top two or three recommendations, each with a rough scope and indicative timeline
  • A readiness score across six weighted dimensions, benchmarked where a real external benchmark exists - and stated plainly where one does not
  • An explicit determination per use case, including any judged to be a broken-process problem rather than an AI problem
  • A directional 90-day plan and a conditional 12-month view
  • The written report, and a live walkthrough of it

Not included

  • Target architecture design
  • Formal vendor or tooling evaluation
  • A board-grade financial model
  • Execution-grade sequencing with named owners
  • Prototypes or demonstrations
  • Fixing the data problems we find
  • Auditing sites or departments not named in the proposal

The audit is genuinely capable of concluding that no AI work is warranted, and sometimes does. That conclusion is worth the fee: it is far cheaper than discovering the same thing eighteen months into a build. An audit that always recommends more work from the same consultant is a sales document, and sophisticated buyers read it as one.

03

Strategy & Roadmap

Stage 2 · Specify · 6-10 weeks

Takes the audit's ranking as its starting point and does not re-open it. What it adds is depth of specification. The practical test: hand an audit report to an engineering team and say build this, and they come back with a list of questions. This is the document that answers those questions in advance - a team could start from it.

What you receive

  • Use cases defined down to triggers, inputs, outputs, decision boundaries and human-in-the-loop points
  • A target architecture an engineer could implement from
  • A reasoned build-versus-buy call on each major component
  • Every data gap converted into scoped, owned, dated remediation work
  • A business case with visible arithmetic, a downside case and an assumptions register
  • A phased roadmap with entry and exit criteria and named owners on both sides

Not included

  • Production code or model training for deployment
  • System integration
  • Executing the data remediation
  • Infrastructure provisioning
  • Running a tender, or negotiating with vendors on your behalf
  • Demonstration prototypes
  • Legal opinion
04

Implementation

Stage 3 · Build · 3-12 months, phased

Implementation of what the roadmap specified: pipelines, model or decision components, integration into your named systems in both directions, deployment into your environment, and testing that includes edge cases, failure modes and user acceptance with the people who will actually use it, on their own documents and their real workload.

What you receive

  • The system built, integrated, tested and deployed into your environment - on-premises, sovereign cloud or your own tenancy
  • Monitoring instrumented against the baseline the audit established
  • The affected workflow redesigned, with escalation and override paths agreed
  • A documented answer to what happens when the model is wrong
  • The people who do the work, and their engineers, trained; the full documentation set
  • A supervised period in which your team operates it and we observe, then an outcome report against the baseline

Not included

  • Indefinite monitoring and operations
  • Round-the-clock support
  • New use cases outside the roadmap design
  • Data remediation beyond what the system requires
  • Hardware or sensor procurement and installation
  • Integration with systems not named in the design
  • Perpetual retraining
05

Governance & Advisory

Stage 4 · Oversee · Ongoing, reviewed annually

A recurring senior oversight engagement over AI systems already in production. It is a thinking and reviewing engagement: your team runs, fixes and changes the systems, and we tell you what the evidence says, what it means, and what to do about it - including when something has crossed the line from advice into work that needs its own engagement.

What you receive

  • Monthly performance and drift reporting against baseline
  • A maintained system inventory with risk classification
  • Incident triage and remediation advice
  • Regulatory-change monitoring scoped to your jurisdiction and sector
  • Quarterly executive and board reporting
  • Custodianship of the roadmap, so new candidate use cases get routed rather than accumulating

Not included

  • Writing production code of any kind
  • Building or retraining models
  • Fixing pipelines or operating your systems hands-on
  • On-call or out-of-hours production support
06

Where you can enter, and where you can stop

Stopping is a normal outcome, not a failure.

No obligation to build with us.The audit stands alone. Its recommendations are yours whether or not you engage us afterwards, and we would rather write an honest report you act on with someone else than a persuasive one you act on with us.
You do not have to start at the audit.Where equivalent work already exists - a credible internal prioritisation, an assessment from another party, an approved design - we will read it, tell you plainly whether it is sound enough to build on, and start from wherever it genuinely leaves off rather than re-charging you for analysis you already own.
Governance can be bought on its own.Oversight does not require that we built the system. It applies equally to AI already deployed by an internal team or another vendor, provided the systems are identifiable, real visibility into metrics and incident history is granted, and someone senior owns the risk.
The sequence is the product.Skipping the diagnosis to reach the build faster is the single most reliable way to end up with a working system nobody needed - which is the failure mode this whole method exists to prevent.
07

What the fee is actually paying for

Thirteen phases spanning all four stages, each with its own communication discipline. A one-off build has none of the nine shaded rows.

When someone hears “AI consultant” they often picture a single deliverable produced by one person typing for a few days. The difference between that and a structured engagement is not effort, it is the phases where your input is formally captured and the work can still change direction cheaply.

All thirteen phases, across the whole relationship from the first diagnostic to ongoing oversight - not one tier.

  • A structured engagement: all 13
  • A one-off build: the 9 shaded rows are missing
  1. 1 First Contact 1-3 days Engagement only
  2. 2 Discovery Call 1 session Engagement only
  3. 3 Recap & Internal Analysis Same day - 3 days Engagement only
  4. 4 Proposal 2-5 days to prepare Engagement and one-off build
  5. 5 Proposal Review & Negotiation 3-10 days Engagement and one-off build
  6. 6 Contracting 1-5 days Engagement only
  7. 7 Onboarding & Kickoff First 5-7 days Engagement only
  8. 8 Delivery Phase(s) Weeks to months, phased Engagement and one-off build
  9. 9 Milestone Reviews Per phase Engagement only
  10. 10 Change Management (as needed) As triggered Engagement only
  11. 11 Final Delivery & Acceptance 1 session Engagement and one-off build
  12. 12 Close-Out 1 session Engagement only
  13. 13 Post-Engagement Ongoing Engagement only
08

Who this is not for

Published, so nobody spends a call finding out - and so we do not have to decline in week two.

Single-site operations with no real back office. If the paperwork is handled by three people who all sit in one room, the honest answer is that better habits will beat a system.
Pure trading businesses with no operations of their own. What we work on is the volume an operation generates - the approvals, the checks, the enquiries, the records - and a buy-and-resell model generates far less of it.
Organisations with no system of record. Without an ERP, a CMMS or something equivalent there is nothing to validate against, and the first project would be that system, not AI.
Anyone who needs a demonstration this month. A diagnostic takes two to four weeks and we would rather lose the work than compress it into a slide deck.

Start here

Start with the audit. Two to four weeks, and you can stop after it.

Tell us the documents, the systems they sit in and the monthly volume, and you will get a scoped answer rather than a brochure.

You get a reply within one working day, from the engineer who would do the work - not a sales sequence.