Hire Data Expert

AI capability assessment

Understand what it will take to adopt and scale AI responsibly

Assess ten connected capabilities, identify blockers and dependencies, compare current readiness with your target, and turn the result into a phased 180-day action plan.

No external APIs or third-party libraries are used. Results depend on the completeness and accuracy of the information supplied.

How it works

Complete the assessment with evidence from relevant teams. Unknown answers are retained as uncertainty and reduce confidence rather than being scored as failure.

1

Rate current capability

Answer thirty weighted questions across the ten capabilities needed for responsible AI delivery.

2

Review readiness and confidence

See your weighted score, target gap, capability profile, confidence level, blockers, and dependencies.

3

Prioritise action

Use the phased recommendations to plan immediate controls, 90-day improvements, and scaling foundations.

AI readiness assessment

Scale: 0 = not in place, 1 = initial, 2 = partly established, 3 = consistently established, 4 = scaled and measured.

0 of 30 questions answered

Organisation name (optional)

Used only in the on-page result and local export.

Target readiness score

Choose 40–100. A typical scaling target is 75.

Strategy and leadership · 13% weight

Rate observable organisational practice, not aspiration.

Executive sponsorship for AI is defined and active.

Select “Not known” when evidence is unavailable.

AI priorities are tied to measurable business outcomes.

Select “Not known” when evidence is unavailable.

Investment principles and target maturity are documented.

Select “Not known” when evidence is unavailable.

Use-case portfolio · 11% weight

Rate observable organisational practice, not aspiration.

AI use cases are assessed using value, feasibility, risk, and data criteria.

Select “Not known” when evidence is unavailable.

A balanced portfolio includes quick wins and scalable strategic initiatives.

Select “Not known” when evidence is unavailable.

Use cases have accountable business owners and benefit measures.

Select “Not known” when evidence is unavailable.

Data readiness · 14% weight

Rate observable organisational practice, not aspiration.

Critical data is discoverable, accessible, and sufficiently reliable.

Select “Not known” when evidence is unavailable.

Data ownership, lineage, quality controls, and permitted-use rules are defined.

Select “Not known” when evidence is unavailable.

Training, evaluation, and retrieval data can be governed through its lifecycle.

Select “Not known” when evidence is unavailable.

Technology and architecture · 11% weight

Rate observable organisational practice, not aspiration.

The architecture supports secure experimentation and controlled production deployment.

Select “Not known” when evidence is unavailable.

Model, prompt, application, and data components can be monitored and versioned.

Select “Not known” when evidence is unavailable.

Integration patterns, environments, capacity, and cost controls are understood.

Select “Not known” when evidence is unavailable.

Talent and skills · 10% weight

Rate observable organisational practice, not aspiration.

The organisation has sufficient AI, data, product, engineering, and domain skills.

Select “Not known” when evidence is unavailable.

Role-based AI literacy and safe-use training are available.

Select “Not known” when evidence is unavailable.

Build, buy, partner, and workforce plans are aligned to the roadmap.

Select “Not known” when evidence is unavailable.

Operating model · 9% weight

Rate observable organisational practice, not aspiration.

Decision rights and delivery responsibilities are clear across business and technology teams.

Select “Not known” when evidence is unavailable.

Reusable delivery standards and multidisciplinary teams support repeatable execution.

Select “Not known” when evidence is unavailable.

Funding, intake, assurance, and lifecycle ownership are defined.

Select “Not known” when evidence is unavailable.

Governance · 9% weight

Rate observable organisational practice, not aspiration.

AI inventory, approval, documentation, and accountability controls are in place.

Select “Not known” when evidence is unavailable.

Policies distinguish permitted, restricted, and prohibited uses.

Select “Not known” when evidence is unavailable.

Performance, incidents, changes, vendors, and retirement are governed.

Select “Not known” when evidence is unavailable.

Responsible AI · 8% weight

Rate observable organisational practice, not aspiration.

Fairness, transparency, explainability, human oversight, and contestability are addressed.

Select “Not known” when evidence is unavailable.

Risk classification determines proportionate reviews and evidence requirements.

Select “Not known” when evidence is unavailable.

Affected stakeholders and potential harms are considered before deployment.

Select “Not known” when evidence is unavailable.

Security and resilience · 8% weight

Rate observable organisational practice, not aspiration.

AI-specific threats, privacy obligations, and access controls are assessed.

Select “Not known” when evidence is unavailable.

Sensitive data, secrets, models, prompts, and outputs are protected.

Select “Not known” when evidence is unavailable.

Incident response, red-teaming, supplier assurance, and resilience plans cover AI systems.

Select “Not known” when evidence is unavailable.

Change and adoption · 7% weight

Rate observable organisational practice, not aspiration.

Users are involved in design, testing, training, and feedback.

Select “Not known” when evidence is unavailable.

Adoption, process change, communications, and workforce impacts are actively managed.

Select “Not known” when evidence is unavailable.

Usage, satisfaction, productivity, quality, and unintended effects are monitored.

Select “Not known” when evidence is unavailable.

Privacy note: this implementation does not transmit information to external services. Submitted answers are processed by the site server for the current request; CSV and JSON exports are generated locally in your browser. Any future storage should be implemented deliberately using secure server-side controls.

Methodology, interpretation, and limitations

How to use the result

Use the score as a structured management diagnostic. Review the lowest capabilities, examine whether strong scores are supported by evidence, and compare readiness with the risk and complexity of the intended use-case portfolio.

The ten weights total 100%. Data and strategy receive the largest weights because weak business alignment or unreliable data can undermine otherwise capable technology delivery. Governance, responsible AI, and security remain essential gating considerations.

Important limitations

The calculator does not test controls, inspect systems, verify evidence, predict financial return, or determine compliance. Self-assessment bias, incomplete stakeholder input, and different interpretations of maturity can materially affect results.

For important decisions, validate the output through workshops, document review, architecture and security assessment, data profiling, use-case due diligence, and legal or regulatory advice where applicable.

Frequently asked questions

What does the AI readiness score measure?

It measures the maturity of ten connected organisational capabilities required to adopt and scale AI responsibly and consistently.

How is the score calculated?

Each known response is scored from 0 to 4, normalised to 100, averaged within its capability, and combined using the published capability weights.

How is “Not known” handled?

Unknown answers are excluded from the readiness score and lower the confidence percentage, making uncertainty visible rather than silently scoring it as zero.

What score indicates readiness to scale AI?

Scores of 75 or above indicate stronger scaling capability. However, low security, governance, data, or responsible-AI scores can still be material blockers.

Can this calculator replace a formal assessment?

No. It is a structured diagnostic for prioritisation and discussion, not an audit, certification, legal opinion, or regulatory determination.

Who should complete the assessment?

A cross-functional group representing business leadership, data, technology, security, risk, legal, HR, operations, and change should contribute.

How often should readiness be reassessed?

Reassess after significant capability improvements, before moving pilots into production, and at least every six to twelve months during active transformation.

Does the tool transmit organisational data?

No external transmission is implemented. The page processes submitted data on the site server for the request, while CSV and JSON exports are generated locally in the browser.

Why set a target readiness score?

The target makes the gap explicit and helps sequence investment according to the ambition and risk profile of the intended AI portfolio.

Can high readiness guarantee successful AI outcomes?

No. Outcomes also depend on use-case selection, execution quality, data conditions, stakeholder behaviour, market factors, and ongoing oversight.

How should low-confidence results be used?

Treat them as a prompt to gather evidence, assign owners, and validate assumptions before making significant investment or deployment decisions.