AI Development Continuum
5 Pillars · 20 Steps · 4 Phases
Continuum Self-Assessment
For each of the 20 steps, rate your own organisation honestly.
Everything runs on your device. Your answers are saved only in your browser.
Align
Steps 1–5Set direction, rules, safe tools and basic literacy
Set the AI ambition
Has leadership tied AI to 2–3 specific mission outcomes, with a named senior owner accountable for them?
Publish an AI acceptable-use policy
Is there a published, plain-language policy on what AI use is allowed and what data must never go into AI tools?
Provide approved, secure AI tools for all
Do all staff have access to an approved, enterprise-grade AI assistant that does not train on your data?
Build AI literacy for leaders and staff
Have leaders and all staff completed practical, role-relevant AI literacy training?
Map and classify sensitive data
Do you know where your sensitive data lives (student, citizen, research, financial, IP) and how it is classified for AI use?
Experiment
Steps 6–10Choose use cases, manage risk and run measured pilots
Build a prioritised use-case portfolio
Is there a single list of AI use cases, scored by value and risk, from which 3–5 priorities have been chosen?
Set up AI risk assessment and a register
Is every AI system listed in a register with an owner, data used and a risk tier, with high-risk uses reviewed before launch?
Launch an AI champions network
Is there an active network of AI champions across units, with protected time to help colleagues?
Run 3–5 measured pilots
Are your pilots measured against a baseline recorded before AI was introduced, with clear success criteria?
Fix data and vet vendors for priority use cases
For each priority use case, is the data fit for purpose and has the vendor passed security, privacy and contract due diligence?
Scale
Steps 11–15Fund what works, redesign work and harden controls
Fund what works: stop or scale
Do you use evidence-based gates to stop underperforming AI initiatives and move funding to those that beat baseline?
Form an AI governance board and map regulations
Is there a cross-functional AI governance group that has mapped which AI laws apply in each place you operate?
Redesign whole workflows
Have you redesigned at least one end-to-end workflow around AI, removing steps rather than just speeding up tasks?
Redesign roles, skills and incentives
Have affected roles, career paths and incentives been redesigned, including a plan for entry-level development?
Put AI security controls in place
Do AI systems run with least-privilege access, activity logging, and an AI-specific incident response plan?
Transform
Steps 16–20Integrate agents, build advantage and reinvent the model
Deploy integrated AI agents with human oversight
Where AI agents act in core systems, are their tasks bounded, with human approval before they send, pay, publish or change records?
Monitor, audit and report transparently
Are AI systems regularly audited for accuracy and fairness, and do you disclose AI use to the people affected?
Turn proprietary data into advantage
Are you deliberately curating your unique data and knowledge so AI gives you an advantage others can't buy?
Embed experimentation in hiring and performance
Is thoughtful AI use part of hiring, performance conversations and leadership expectations?
Measure value and rethink the operating model
Does leadership review AI value every year and redesign structures, services and the operating model accordingly?