Home / Decisions

Decide before you build.

Most AI money is lost in the decision, not the deployment. These products make the call defensible — before a dollar of hardware or development is committed.

Fixed price, set in a one-page scope before work starts. No time-and-materials, and the prices are published on purpose.
Free · 3 minutes · no call

Not sure which of these you need?

Almost nobody arrives knowing. The Readiness Score rates you across five dimensions and names the weakest one — and that weak spot is what tells you which decision to buy next, or that you shouldn’t buy one yet.

Take the Readiness Score →
Strategy & use cases Data People & ownership Processes Execution track record

AI Adoption Decision

A sequence, not a menu — each step ends in something signed, and each takes the previous one as its input. Most companies don’t need all three, so every step below says when to skip it. Doing them in order also makes each one faster: a blueprint after the program takes days rather than weeks, because its input is already collected.

01
Should we start at all?

AI Adoption Assessment

An expert review of where AI creates measurable value in your business, and the executive decisions needed to get there.

From $20,000 · from 3 weeks
Ends in: a company-level verdict, up to and including “not yet.”
Skip it if the initiative is already chosen and leadership agrees on it.
02
What, and in what order?

AI Adoption Program

Greenlit opportunities turned into a prioritized 12–24 month portfolio, with direction and investment outlook per initiative.

From $30,000, from 2 weeks · with the assessment, from $48,000
Ends in: which initiatives, in what order, decided one by one.
Skip it if you have one initiative rather than a portfolio — go straight to the blueprint.
03
How exactly is it built?

AI Solution Blueprint

The full functional and technical design a delivery team builds from — models, agents, integrations, security, acceptance criteria. Vendor-neutral. Runs standalone for a single system, or per initiative across a portfolio.

One system from $30,000 · 5–6 weeks, no prior stage · portfolio from $15,000 per initiative · credited against the build
Ends in: an approved design that becomes the build baseline.
Skip it only if you already have a specification a build team can quote from.

AI Compute Decision

Own or rent, exit the cloud or stay, audit what you already run. Not a sequence — these are independent calls, and you take the one that’s actually on your table.

Free · no call required
Want a number before you talk to anyone? Three validated reference architectures with indicative pricing — an inference pod, a fine-tuning and training cluster, and a starter stack from $95K. Budget from those before committing to anything.
See reference architectures →
What are we running today?

Infrastructure Audit

An honest snapshot of what you actually have — true cost, real load, where it idles and where you’re locked in. The cheapest way to stop guessing before spending on more.

From $5,000 · timeline set in the scope
Ends in: the current state in numbers — including what you can stop paying for.
Repatriate this workload, or stay?

Cloud Exit Assessment

A reasoned answer for one rented workload: an honest break-even and the cost of being stuck either way, on one page for the board. Sometimes the answer is “stay,” and we say so plainly.

Fixed price · timeline set in the same one-page scope
Ends in: a Go / Conditional-Go / No-Go verdict you can defend to a board.

Also on this track, not open yet: Investment Review — an independent second opinion on a major compute investment before you sign · Sovereign Stack — an AI stack that keeps data, models and control on infrastructure you own.

Budget, but no AI team

Use ours.

A senior AI team for the length of one decision. No hire, no headcount, no bench.

10engineers on AI full time
4–10on your engagement
2017building since
The architect who writes your design stays on it — and builds it if you go ahead. Remote, on Asia and Gulf hours.
Who you deal with
Technical leadThe architecture and every technical call.
Product leadScope, schedule, and your one point of contact.
AI/ML engineersThe people who would build it.
Infrastructure engineersWhen the decision is about compute.

Why teams pay for the decision

We’re not selling you the answerThe blueprint is written to be executable by any competent team, including one that isn’t us. A decision you can take to another vendor is the only kind worth paying for.
“No” is a valid outputSometimes the verdict is stay in the cloud, or don’t start yet. We say it plainly, and it’s the cheapest sentence you’ll ever buy from us.
A document, not a slideWhat you get is written to survive a board meeting and a legal review — a verdict with the conditions it depends on, not a deck that impresses a room and then evaporates.
What we need from youAccess to four to six stakeholders for short interviews, the relevant documents, and one person to coordinate. That’s the whole ask — most of the discovery happens on our side.

What a wrong start actually costs

Not the failed pilot’s budget. That’s the smallest line on the bill, and the only one anyone writes down.

DirectThe budget spent on a pilot that went nowhere. Visible, precise, and the least of it — which is exactly why it distracts from the other two.
OpportunityTwo quarters of your most capable people spent proving the wrong use case can’t work — instead of shipping the winnable one. Talent is the scarce resource, not cash.
CredibilityThe lesson a visible failure teaches the board: “AI doesn’t work here.” The next initiative gets funded slowly, if at all. Largest, compounds, and never appears in the post-mortem.

The cost of a wrong AI start →

Decision made? Then it gets built.

The same team that helped you decide takes it to a system that runs — the AI software, physical AI, or the infrastructure under it. That’s the point of deciding properly first: the build starts from an approved design instead of an argument.

See what we build →

Frequently asked questions

Do we have to do all of them, in order?

No, and most companies don’t. The adoption products chain when you need the chain — but if the initiative is already agreed, the blueprint alone is a complete piece of work. Each card above says when to skip it.

How long does each one take?

Adoption Assessment from three weeks, Adoption Program from two, Solution Blueprint from seven days. The blueprint varies most, because it depends on the size of the project and on whether the assessment and program have already been done — their output is its input. For the Infrastructure Audit and Cloud Exit Assessment the timeline is set alongside the price, in the one-page scope.

What do you need from us?

Access to four to six stakeholders for short interviews, the relevant documents, and one person to coordinate scheduling. That’s the whole ask — most of the discovery happens on our side.

Do you need access to our systems, data or source code?

Not for the Adoption Assessment — it runs without access to your corporate network, databases or source code. The infrastructure products need more: the audit works from billing, utilisation and architecture documentation, and exactly what’s required is agreed in the scope before anything starts.

Can we skip straight to the blueprint?

Yes, if the initiative is already chosen and leadership agrees on it — that’s the most common shortcut and a legitimate one. What you can’t skip is having a design before a build; that’s the whole argument of this page.

What if the answer comes back “no”?

Then you have it in writing, with the reasoning, for the price of the assessment rather than the price of a build. That’s the product working, not failing — a verdict of “not yet” or “stay in the cloud” is the outcome we’re paid to reach when it’s the true one.

Fixed price, or time and materials?

Fixed. Scope, timeline and price are set in a one-page scope before anyone starts. If the scope changes mid-engagement we re-quote in writing first, before any extra work happens.

Can we take the output to another vendor?

Yes — the blueprint is deliberately written to be executable by any competent delivery team. If it only worked with us, it wouldn’t be a specification.

We already have an assessment from someone else. Does it count?

Bring it. We’ll read it and tell you what it settles and what it leaves open before quoting anything — sometimes it’s enough to go straight to the blueprint, and sometimes the gaps are the expensive part.

What happens once a decision is made?

It gets built — by us, your team, or a third party, since the design is portable by construction. See what we build →

Still not sure?

Tell us the decision on the table and we’ll point you to the right product — including when the answer is none of them.