Skip to content
Assessment · AI Readiness

AI Readiness Assessment

Most artificial intelligence pilots never reach production. Not because of the technology, but because usable data, governance or a business case to sustain the investment are missing. We evaluate how ready your organisation is to take AI to production and build the roadmap to get there.

What it is
A pilot that works is not a capability that produces.

The assessment reviews eight fronts that determine whether an AI case can move from demonstration to operation: from data quality to who answers when the model gets it wrong. It turns the gaps into a roadmap prioritised by business value and agrees it with the areas. It does not evaluate tools: it evaluates the ability to reach production.

01

Know which cases are viable

Which of your ideas have the data, process and value to justify the investment.

02

Close the right gaps

What exactly is missing — data, platform, talent or governance — before scaling.

03

Invest with a business case

Each initiative with the value it returns and the effort it demands, not with a promise.

Assessment or consulting

One surveys and maps the route. The other executes it.

These are two different services and it is worth knowing which one you need before contracting. The assessment does the first information-gathering exercise, builds the roadmap and aligns it with the business areas. The consulting practice executes that roadmap.

AI Readiness Assessment — this page

Surveys, maps and aligns. It surveys candidate use cases, data maturity, platform, talent and governance; measures the gap against recognised frameworks; builds the adoption roadmap and agrees it with the areas. It runs two to four weeks depending on the size of the organisation. It does not implement, does not configure and does not operate anything.

Artificial Intelligence — execution

Executes the roadmap. It takes the roadmap into production: solution design, data engineering, model implementation, integration with the processes and continuous operation. It works just as well if you already have a roadmap of your own.

What it covers

Eight fronts that decide whether AI reaches production.

An AI case fails at its weakest link, not at its average. We review each front with the same depth and rate it by maturity level.

01

Strategy and use cases

Which business problems you want to solve with AI and which of them justify the investment against simpler alternatives.

02

Available data

What data exists, where it lives, at what quality, and whether it can be trusted to train or feed a model.

03

Platform and infrastructure

What compute, storage and integration capacity you have today and what would be needed to run in production.

04

Talent and skills

What the team can do, what is outsourced, and where a critical dependency on one person sits.

05

Governance and ethics

Who approves a case, who answers for an automated decision, and what responsible-use criteria are defined.

06

Security and privacy

Which personal or sensitive data is involved, how it is protected and what the applicable regulation requires.

07

Process and adoption

Whether the process AI will touch is documented, and whether the people who run it are ready for the change.

08

Value measurement

How the return will be measured and who answers for it once the case is operating.

How we do it

Six phases, without interrupting your operation.

The assessment rests on interviews, review of data sources and documentary analysis. It does not train models or intervene in your systems.

01

Scope and preparation

We define which areas, processes and data sources are in, and agree read-only access.

02

Business interviews

With the areas proposing use cases, to understand the real problem behind each idea.

03

Data and platform review

Available sources, quality, governance and technical capacity, checked against what the interviews said.

04

Gap analysis

Comparison against NIST AI RMF and ISO/IEC 42001, front by front.

05

Use-case prioritisation

Each candidate with viability, effort and expected value, to decide where to start.

06

Business alignment

Presentation of the roadmap with leadership and the areas, so priorities are agreed.

Method and frameworks

What your maturity is measured against.

The evaluation does not rest on the judgement of whichever consultant shows up, nor on a vendor catalogue, but on public and auditable frameworks your team can consult and your auditor will recognise. Those frameworks sit alongside the law that decides what data a model may use: Habeas Data (Ley 1581) in Colombia and the LFPDPPP in Mexico.

01

NIST AI RMF 1.0

Evaluation framework. The AI risk management framework: govern, map, measure and manage. It orders the maturity rating.

02

ISO/IEC 42001

Evaluation framework. The AI management system standard. It is the structure an audit or a corporate client will ask for.

03

ISO/IEC 23894

Risk guide. AI-specific risk management: bias, explainability, model drift and data dependency.

04

OECD AI Principles

Responsible-use guide. International reference for transparency, fairness and accountability in automated systems.

05

EU AI Act

Regulatory guide. Classification by risk level. It applies if you operate or sell in Europe, and it sets the regional regulatory direction.

06

MLOps maturity

Operational guide. What separates a model in a notebook from a model in production with monitoring, versioning and retraining.

Risks of going without

What it costs to scale before you are ready.

These risks do not show up in the pilot. They show up the day the case reaches production and touches a real customer.

01

Pilots that do not scale

The demo works on hand-prepared data and collapses when it meets the real data of everyday operation.

02

Insufficient data

Halfway through the project you discover the information needed does not exist, is incomplete or cannot be used.

03

Decisions with no owner

The model gets it wrong and nobody defined who answers, how it is reviewed or how it is corrected.

04

Regulatory exposure

Personal data or automated decisions with no legal basis and no traceability, in a tightening regulatory environment.

05

Spend with no return

Platform and licences get bought before knowing which case justifies them.

06

Vendor dependency

Knowledge stays outside and changing platform means redoing the work from scratch.

What it asks of your team

What it costs you in time, said upfront.

An assessment that does not state the commitment it requires ends up delayed. This is what we need from your side to deliver on time.

01

Two to four weeks

Two weeks in organisations of up to 50 employees; four between 51 and 300. The timeline is agreed before starting.

02

Scheduled interviews

Sessions with the areas proposing use cases, with IT and with whoever answers for the data.

03

Read-only access

Queries against the platforms and existing documentation. At no point is a configuration modified.

04

A single point of contact

One person coordinating schedules and access. It is the factor that most affects hitting the deadline.

05

Whatever documentation exists

Inventory of data sources, privacy policies, current architecture and results of previous pilots.

06

A closing session

The presentation of findings with leadership and the areas involved, where the roadmap priorities are agreed.

Who it is for

When it makes sense and when it does not.

It makes sense if…

Your organisation has pilots that never quite scale; you are about to set an AI budget and need to justify it; you handle personal data and want to know your exposure; or you want to know which of your ideas are viable before hiring anyone.

Probably not if…

You already have the case defined, the data ready and what you need is to build it: go straight to the Artificial Intelligence practice. Or if the underlying problem is the quality of your data: that is measured by the Data & Analytics Maturity Assessment.

Benefits

What you gain from the assessment.

Cases prioritised by viability

Which ones have the data, process and value, and which are better discarded before investing.

Explicit gaps

What exactly is missing in data, platform, talent and governance before you can produce.

Regulatory risk contained

What the applicable regulation demands of the cases you want to run, before you are exposed.

Investment backed by figures

An economic case per initiative, with effort and expected value.

Defined governance

Who approves, who answers and how an automated decision gets reviewed.

A comparable baseline

A starting point that lets you measure progress next year on the same criteria.

The SUMāTO approach

Why this evaluation and not a proof of concept.

The difference is not in showing the technology works: it is in knowing whether your organisation can sustain it in production.

01

Evaluation, not demonstration

A pilot proves the technology. This assessment evaluates whether there is data, process, governance and talent to operate it.

02

Recognised frameworks

The comparison is against NIST AI RMF and ISO/IEC 42001, not against whichever consultant shows up.

03

Business language

Each gap with its cost of inaction and the use case it blocks.

04

Vendor independence

The roadmap is not shaped by the platform that would suit us to sell afterwards.

05

No interruption to operations

Interviews and document review. We do not train models or touch your systems.

06

Continuity into execution

If you decide to proceed, the roadmap connects with the Artificial Intelligence practice without starting the survey again.

The conclusion

As-Is, To-Be and the plan to get from one to the other.

Every assessment closes with the same structure, whatever the practice: where you stand today, where you need to be, what separates the two states and in what order that distance gets closed.

01

Current state — As-Is

The starting point surveyed with evidence, not declared in an interview: what exists, how it operates and how far it sits from what the business needs.

02

Target state — To-Be

Where the organisation needs to get to, defined with the business areas rather than imposed by the consultant. It is the benchmark everything else is measured against.

03

Gap analysis

Every difference between the As-Is and the To-Be, with everything required to close it: technology, processes, people, governance and budget. No gap is stated without what it demands.

04

Risk matrix

Each gap rated by probability and business impact, so priority does not depend on who pushes hardest but on what it costs to leave it open.

05

Work plan

The concrete sequence to reach the To-Be: what comes first, what it depends on, how much effort it takes and who should answer for each front.

06

Business alignment

The plan is presented and agreed with the areas involved. A roadmap signed only by IT does not survive the first quarter.

The report

How what you receive is structured.

The central deliverable is a report with a fixed structure, designed so leadership reads the first pages and the technical team works with the rest.

01

Executive summary

Two pages: readiness level, the three gaps that weigh most and what decision each one calls for.

02

Maturity by front

Rating of the eight fronts with the gap made explicit against NIST AI RMF and ISO/IEC 42001.

03

Use-case inventory

Each candidate with viability, data required, effort and expected value.

04

Data diagnosis

Which sources exist, at what quality, and what is missing to sustain the prioritised cases.

05

Adoption roadmap

Prioritised initiatives, with the gaps to close before each one.

06

Immediate actions

What can be corrected or started without a project or additional budget.

Deliverables

What you receive at the end.

  • Current state (As-Is): readiness to take AI to production, rated across each of the eight fronts against NIST AI RMF and ISO/IEC 42001.
  • Inventory of candidate use cases, prioritised by feasibility and value.
  • Diagnosis of the data sources that underpin the prioritised cases.
  • Target state (To-Be): the data, governance and operating conditions the first production case demands, agreed with the business.
  • Gap analysis between the As-Is and the To-Be, with each gap, its evidence and everything required to close it: data, platform, governance and capabilities.
  • Risk matrix: regulatory, privacy and model exposure, rated by probability and business impact.
  • Work plan to reach the To-Be: an adoption roadmap with the gaps that must close before each initiative.
  • Governance recommendations: who approves, who answers and how it gets reviewed.
  • Immediate-impact actions, executable without additional budget.
  • Executive presentation for committee and leadership.
  • Alignment session with the business areas involved.
Frequently asked questions

About the AI Readiness Assessment.

What exactly do you evaluate?+
Eight fronts: strategy and use cases, available data, platform and infrastructure, talent, governance and ethics, security and privacy, process and adoption, and value measurement. Each is rated for maturity against recognised frameworks.
How does it differ from AI consulting?+
The assessment does the first information-gathering exercise, builds the roadmap and aligns it with the business areas. The Artificial Intelligence practice is the one that executes that roadmap: designs the solution, builds the models and takes them to production.
Do we need data ready to do this?+
No. One of the fronts evaluated is precisely what data exists and at what quality. If the diagnosis concludes the data does not support the cases, that is a valid result and it prevents an investment that would have failed.
Do you train any model during the assessment?+
No. It is a capability evaluation: interviews, source review and documentary analysis. Nothing is trained, nothing is deployed and no production system is touched.
Who should take part on our side?+
The areas proposing use cases, the IT team and whoever answers for the data. Without the business at the table, prioritisation ends up technical rather than value-driven.
Is it useful if we already have pilots running?+
Yes, and it is usually the best moment: the assessment explains why those pilots do not scale and what needs closing for them to reach production.
Does it cover regulatory compliance?+
It evaluates exposure: which personal data is involved, which decisions are automated and what the applicable regulation requires. It does not replace legal advice, but it shows where advice is needed.
How often should we repeat it?+
Annually, or whenever the case portfolio changes substantially: data and platform maturity move quickly once initiatives are under way.
The first step

Know what you are missing before investing in artificial intelligence.

Book your AI Readiness Assessment and get an adoption roadmap prioritised by value, with the data, platform, talent and governance gaps made explicit.

Book my assessment See the Artificial Intelligence