AI Transformation Framework
A three-phase, 10-pillar playbook for SMBs of 10-100 people, where change is personal, budgets are real, and every initiative competes for the same people's attention.
Book a Discovery CallBuilt for SMBs running real client work
Software houses, digital agencies, and product-led teams between 20 and 200 people. Leaders who can't pause delivery to 'do AI', but know that ignoring it is starting to cost them. The framework treats AI as augmentation of people and process, not replacement - and it's designed to run alongside live client work, not instead of it.
Three phases, each building on the last
Assessment
Understand where you are, what you have, what's blocking you.
Map current state across 10 dimensions to determine readiness, identify quick wins, and surface hidden blockers before investing in strategy.
Ideation & Plan
Brainstorm through two lenses, prioritise, define initiatives, build a quarterly plan.
Translate assessment findings into concrete AI initiatives with owners, timeboxes, and stakeholder alignment.
Change Management
Scale what works, drop what doesn't, build muscle memory.
Keep the transformation alive after the initial excitement fades - handle resistance, track adoption, run quarterly retros.
10 pillars → Readiness Profile
The Assessment phase produces a Readiness Profile: a per-pillar judgement on a simple traffic light. Below are the pillars, the top questions we ask, and the signals we look for in each.
Blocker - must be addressed before AI initiatives can start
Risk - gaps exist but can be worked around in parallel
Ready - sufficient foundation to proceed
P01Technology Awareness
Does the organisation have the technical foundation and literacy to adopt AI tools?
Technology Awareness
Does the organisation have the technical foundation and literacy to adopt AI tools?
- 1.How technical is the company overall? (Engineering-heavy vs. business-heavy vs. mixed)
- 2.How technical are the employees who will be primary AI users?
- 3.What is the current level of automation maturity? (Manual > scripts > CI/CD > AI-assisted)
- 4.Do teams understand the difference between AI-assisted work, AI agents, and full automation?
- 5.Is there an internal champion or group already experimenting with AI tools informally?
No one uses AI tools casually; leadership views AI as 'hype'; engineering is resistant to AI-generated code.
Multiple people already use ChatGPT, Claude or Copilot; leadership asks 'how can we do this with AI?' regularly.
P02Knowledge Base & Data Readiness
Is the organisation's data in a state where AI can actually use it?
Knowledge Base & Data Readiness
Is the organisation's data in a state where AI can actually use it?
- 1.Do we have structured data for AI agents? (project templates, client briefs, process docs)
- 2.How confident are we in data cleanliness and consistency?
- 3.Where does institutional knowledge live? (people's heads vs. wikis vs. scattered docs)
- 4.Is documentation current, or a graveyard of outdated SOPs?
- 5.What percentage of workflows are documented well enough for an AI agent to follow?
Knowledge lives in three people's heads; documentation is two-plus years old; no naming conventions.
Active wiki or knowledge base; templates exist for common workflows; data mostly in one system.
P03Security
Can the organisation use AI tools safely without exposing sensitive data?
Security
Can the organisation use AI tools safely without exposing sensitive data?
- 1.Can we use our data with external AI services? (data classification policy)
- 2.What data categories should never go into AI tools? (Client IP, PII, financials, credentials)
- 3.How do we secure selected AI tools? (SSO, audit logs, data residency, retention policies)
- 4.Do we have a policy for what employees can and cannot input into AI tools?
- 5.Who is responsible for reviewing AI security posture?
No data classification policy; employees paste client code into free-tier AI tools; no audit trail.
Existing security policies extendable to AI; data classification in place; InfoSec team engaged.
P04Legal & Compliance
Are there legal or contractual barriers to AI adoption?
Legal & Compliance
Are there legal or contractual barriers to AI adoption?
- 1.Do client contracts restrict AI usage on their projects?
- 2.What sensitive data do we handle with legal implications? (Offers, salaries, resumes, health data)
- 3.Are there industry-specific compliance requirements? (GDPR, SOC 2, HIPAA, ISO 27001)
- 4.Who reviews AI tool vendor agreements for data processing terms?
- 5.What's our IP position - do we own AI-generated outputs?
No one has read client contracts for AI clauses; using AI on regulated data without review.
Legal has reviewed key contracts; clients are informed; vendor agreements are vetted.
P05People & Culture
How ready are your people for change, and where is the resistance?
People & Culture
How ready are your people for change, and where is the resistance?
- 1.How open are employees to change and uncertainty?
- 2.Who are the AI champions - the people already experimenting and excited?
- 3.Who are the sceptics, and what specifically are they worried about?
- 4.What's the organisation's track record with past technology changes?
- 5.Is there psychological safety to experiment and fail?
Past changes were forced and poorly communicated; middle management is resistant; fear of job replacement is widespread.
People sharing AI tips in Slack; leadership models AI usage; past changes went relatively smoothly.
P06Budget & Investment Capacity
Can the organisation afford to invest, and for how long?
Budget & Investment Capacity
Can the organisation afford to invest, and for how long?
- 1.Is there a dedicated budget for AI transformation?
- 2.What's the monthly or annual budget for AI tool subscriptions?
- 3.Is there budget for training and transition time?
- 4.What's the expected ROI timeline - three months or twelve?
- 5.Have we calculated the hidden costs? (Integration, security reviews, process redesign)
No dedicated budget; expectation of immediate ROI; training time not factored in.
Allocated innovation budget; leadership understands the J-curve; willing to invest in tooling.
P07Tools & Technology Stack
What's the current tool landscape, and what can AI plug into?
Tools & Technology Stack
What's the current tool landscape, and what can AI plug into?
- 1.What tools are currently used across the organisation?
- 2.Who is using what - and are there tool silos between teams?
- 3.Which tools already have AI features we're not using?
- 4.What automation tools are in play? (Make, Zapier, n8n)
- 5.Which tools have APIs that AI agents could integrate with?
Tools are disconnected silos; no automation layer; teams use different tools for the same purpose.
Tools have APIs; an automation platform is already in use; some AI features already enabled.
P08Processes & Workflows
Which workflows are candidates for AI augmentation, and which are too messy to touch?
Processes & Workflows
Which workflows are candidates for AI augmentation, and which are too messy to touch?
- 1.What are the top 10 most time-consuming repeatable workflows?
- 2.Which are documented enough that someone new could follow them?
- 3.Where do people spend time on tasks that feel like 'AI should do this'?
- 4.Are there bottleneck processes where one person blocks everyone else?
- 5.What's the 'groan test' - which tasks make people groan?
No process documentation; every project reinvents the wheel; processes too tangled to describe.
Some workflows already templated; teams can identify their top three time sinks; documentation habits exist.
P09Data Governance
Beyond security: who owns data, how is quality maintained, and what governance structures exist?
Data Governance
Beyond security: who owns data, how is quality maintained, and what governance structures exist?
- 1.Who owns data quality in the organisation?
- 2.What's the data lifecycle - from creation through archival?
- 3.Are there naming conventions, tagging systems, or metadata standards?
- 4.How is access controlled - who can see what?
- 5.Would an AI agent find consistent, trustworthy information in our knowledge base?
No data ownership; everyone structures data differently; no retention policy.
Consistent data standards; someone owns data quality; access controls exist.
P10Readiness, Capacity & Change Management
Does the organisation have the people, bandwidth, and change muscle to execute transformation?
Readiness, Capacity & Change Management
Does the organisation have the people, bandwidth, and change muscle to execute transformation?
- 1.Do we have people who can lead the transition?
- 2.Do we have a budget for innovation, or do we need to carve time from delivery?
- 3.What's the current utilisation rate - is there bandwidth?
- 4.Can we protect initiative time from being cannibalised by urgent client work?
- 5.What transformation initiatives have been successfully executed in the past?
100% utilisation; no internal champion; every initiative gets deprioritised for client work; change fatigue is real.
Someone allocated to lead; 10-20% time protected; organisation has a 'we try things' culture; past changes stuck.
The Readiness Profile
After interviewing stakeholders across all 10 pillars, you get one synthesised artefact - not a numerical score, but a professional judgement per pillar with a key finding and a priority action.
- Any Red pillar = blocker. Address before AI initiatives can start there.
- Mostly Amber = discovery mode. Start with the highest-readiness areas.
- Mostly Green = ready for systematic rollout.
From findings to a quarterly plan
Phase 2 turns the Readiness Profile into a prioritised set of AI initiatives with owners, timeboxes, and a four-quarter roadmap. Two to four weeks.
Step 1 - Ideation through two lenses
Run a 60-90 minute brainstorm with the initiative candidates from Assessment. Aim for 10-20 ideas across both lenses.
Project Level
“Where could AI improve how we deliver specific client projects?”
Focus on workflows within a single project that are repetitive, time-consuming, or error-prone.
- Weekly cadence bottlenecks
- Recurring quality issues
- Client-visible improvements
- Team frustrations
Team / Department Level
“Where could AI change how an entire team or department operates?”
Focus on cross-project patterns - workflows every project team does, or department-wide processes.
- Repeatable DM tasks
- HR screening time
- Marketing repetition
- Knowledge-sharing gaps
Step 2 - Prioritisation matrix
Place every brainstormed idea on a Pain × Readiness matrix. Only the top-right quadrant - High Pain, High Readiness - becomes a starting initiative this quarter.
Next-quarter roadmap. Build readiness now.
1-3 max, 4-6 week timebox. Your first initiatives.
Park it. Revisit if the landscape changes.
Run in parallel. Opportunistic, low-risk.
Pain = time, money, or frustration cost. Readiness = team willing, data ready, tools available.
Step 3 - From idea to defined initiative
T01Initiative definition
Each ‘Start Here’ idea becomes a defined initiative with one owner and timebox of 4-6 weeks.
Initiative definition
Each ‘Start Here’ idea becomes a defined initiative with one owner and timebox of 4-6 weeks.
- Initiative title
- Lens (Project / Team)
- Initiative description
- Team involved
- Workflow to augment
- Current state
- Target state
- Success metrics
- Owner (one person)
- Timebox (4-6 weeks)
- What worked? - Keep and codify into the Plugin Pool.
- What didn't? - Drop or redesign.
- What surprised us? - Investigate and adapt.
- What's next? - Pull the next idea from the matrix.
T02Communication strategy
Every audience hears the right message, through the right channel, at the right cadence. Start before the first initiative launches.
Communication strategy
Every audience hears the right message, through the right channel, at the right cadence. Start before the first initiative launches.
| Audience | Message | Channel | Cadence |
|---|---|---|---|
| Leadership | ROI, strategic positioning, competitive advantage | Monthly exec report | Monthly |
| Initiative teams | Hands-on guidance, quick wins, support | Slack + standups | Weekly |
| All employees | Vision, progress, what's coming, why it matters | All-hands + newsletter | Bi-weekly |
| Sceptics | Address concerns, share evidence, invite to observe | 1:1 conversations | As needed |
| Clients | Transparency about AI usage, quality assurances | Account conversations | Per project |
The AI Transformation Plan
Distribute initiatives across four quarters. Q1 is detailed; Q2 is planned but adjustable; Q3-Q4 are directional. Each initiative has exactly one owner.
Foundation & first initiatives
- Initiative
- Initiative
- Initiative
- Initiative
Validate & expand
- Initiative
- Initiative
- Initiative
Systematise & optimise
- Initiative
- Initiative
Strategic integration
- Initiative
- Initiative
- Review at the end of every quarter. Q1 results reshape Q2, and so on.
- Each initiative has exactly one owner. Shared ownership = no ownership.
- If an initiative slips, decide explicitly: move to next quarter, adjust scope, or drop it.
- New ideas mid-quarter go to the Prioritisation Matrix, not directly onto the plan.
Keep the transformation alive
Phase 3 starts the moment Phase 2 ships and never ends. A quarterly review rhythm, a resistance playbook, and an adoption dashboard keep the plan honest.
Step 1 - Quarterly review rhythm
Four cadences, four conversations. Each one ends with an explicit decision, not a status update.
Initiative retrospective
Initiative team + champion
Lessons learned; decision: scale, adjust, or drop.
Quarterly review
Lead + Sponsor + team leads
Updated Transformation Plan.
Mid-year retrospective
All teams + leadership
Course correction, budget review.
Annual review
Full leadership + all teams
Next year's plan, celebration.
- What worked? - Codify into the Plugin Pool, document as a case study.
- What didn't? - Drop or redesign. Don't carry failed experiments into the next quarter.
- What surprised us? - Investigate and adapt. Surprises are often the most valuable signal.
- What's next? - Pull from the Prioritisation Matrix. Re-prioritise if the landscape has changed.
Step 2 - Resistance management playbook
Five recurring patterns, five interventions. Resistance is a signal, not a problem - it tells you which conversation to have next.
R1Fear of replacement
Fear of replacement
“AI will take my job.”
Show AI as augmentation; highlight how top performers use AI to do more, not less.
R2Quality concerns
Quality concerns
“AI output isn't good enough.”
Demonstrate with real examples; involve sceptics in review so they own the quality bar.
R3Effort concerns
Effort concerns
“Learning this is more work.”
Protect learning time; show quick wins early; pair sceptics with internal champions.
R4Control concerns
Control concerns
“I don't trust AI decisions.”
Keep human in the loop; show audit trails; never promote AI output autonomously.
R5Philosophical
Philosophical
“This isn't how we should work.”
Acknowledge the concern; share evidence over argument; let results speak.
Step 3 - Adoption metrics dashboard
Two lenses on adoption: leading indicators tell you whether people are using the system; lagging indicators tell you whether it's working.
Leading indicators
Weekly- Plugin usage count, installs, invocations
- Number of active Claude Code users
- New plugins or skills created this week
- Support requests and questions
Tells you whether adoption is happening.
Lagging indicators
Monthly- Time saved per workflow (before/after)
- Output quality metrics (defect rates, revision cycles)
- Team satisfaction scores
- Client feedback on AI-augmented deliverables
Tells you whether it's actually working.
The living Transformation Plan
The plan is never finished. Each quarter's results reshape the next quarter's priorities. New ideas mid-quarter go to the Prioritisation Matrix, not directly onto the plan.
Want a Readiness Profile for your org?
Book a 30-minute discovery call. We'll walk you through the 10 pillars against your current setup and tell you whether AI transformation is your next move - or whether something else is.