Head of Digital
AI Transformation Lead
I turn AI from a tool conversation into an operating model — strategy, capability, governance, workflow and measurement. Built inside a 40-person agency, proven on a global automotive account spanning 140+ markets.
40
Weekly active Claude users across the agency (approx.)
5h
RFQ response time, down from 3–5 days
30%
Of my role formally allocated to AI transformation
13+
Roles covered by AI playbooks
How I run it
Five pillars. Select one to filter the work below.
01
Strategy & roadmap
3 initiatives
AI transformation mandate
Role-level AI impact map
AI search visibility (AEO)
02
Governance
3 initiatives
AI governance layer
AI platform due diligence
AI & tool risk checks
03
Workflow redesign
7 initiatives
AI pitch playbook
Vendor master consolidation
Team calendar auto-sync
Request intake automation
Estimate drafting via MCP
AI-built websites, end to end
Full-AI social production
04
Capability building
2 initiatives
Shared Claude Skills library
Claude Playbook & role guides
05
Measurement
2 initiatives
AI adoption baseline
AI production cost benchmark
Use cases
17 initiatives, ordered by evidence: measured → modeled → qualitative.
01
AI pitch playbook
A Claude-based playbook for RFQ and pitch responses.
RFQ response time cut from 3–5 days to about 5 hours, proven on a global automotive account.
Workflow redesign
02
AI transformation mandate
Phased AI roadmap and role-formalization proposal for leadership.
Formally approved, with 30% of my time protected for AI work.
Strategy & roadmap
03
Vendor master consolidation
Merged accounting, project-system and Slack vendor records into one directory.
Reconciled 332 payees against 245 system vendors.
Workflow redesign
04
Team calendar auto-sync
Every staff member’s daily calendar synced into the company hub.
33 staff, refreshed twice daily, zero per-person setup.
Workflow redesign
05
Request intake automation
Client web-form requests become Notion projects with a standard task plan.
Every request auto-creates six dated tasks.
Workflow redesign
06
Shared Claude Skills library
Recurring agency workflows codified as shared Claude Skills.
Five skills live: SoW drafting, message triage, tracker updates, dev briefs, daily brief.
Capability building
07
Claude Playbook & role guides
A 12-lesson curriculum and role-by-role AI guides in Notion.
Company-wide AI curriculum covering 13+ roles.
Capability building
08
Role-level AI impact map
AI task exposure mapped across 15 agency roles.
Used by leadership to shape hiring and role redesign.
Strategy & roadmap
09
AI search visibility (AEO)
Made the agency’s website readable and citable by AI assistants, then packaged the method as a client offer.
Structured data and FAQ live since Sep 2026; extended into a client proposal.
Strategy & roadmap
10
AI adoption baseline
Role-based usage survey feeding one-to-one workflow interviews.
35 staff interviews analysed; survey cut from 18 to 5 questions.
Measurement
11
Estimate drafting via MCP
Claude connected to the agency’s estimate system through a custom MCP server.
Claude drafts projects, billing fields and line items; pilot under way.
Workflow redesign
12
AI-built websites, end to end
Mockups, design, front-end code and deployment produced with Claude, including client web builds.
This portfolio was designed, coded and shipped through Claude + Framer MCP: bilingual EN/JA, CMS, 12 custom code modules.
Workflow redesign
13
AI production cost benchmark
AI tools benchmarked against internal production rates per creative task.
5-language voice-over: −86% cost, ~95% time saved (modeled).
Measurement
14
AI governance layer
Responsible AI guidelines plus confirmed / assumption / needs-verification labels on AI drafts.
Shared rules for every staff member; stops AI assumptions being read as fact.
Governance
15
AI platform due diligence
Evaluated an enterprise creative AI platform for agency-wide adoption: capability, security, IP and commercial terms.
Assessment and vendor negotiation led end to end; decision pending.
Governance
16
AI & tool risk checks
Audited tool permissions, integrations and suspicious messages before they became incidents.
Verified a billing email via DKIM/SPF/DMARC and flagged a Slack file-exposure risk.
Governance
17
Full-AI social production
Client social video and imagery generated entirely with AI tools.
Delivered for a global electronics brand.
Workflow redesign
Where I use AI
Seven areas, from building to governing.
Research
Competitive & market research
Cross-tool fact finding
AI industry briefings
Writing
Bilingual proposals and SoWs
EN/JA stakeholder messaging
Developer specs and bug reports
Creative
AI video and imagery
AI voice-over
Multilingual localisation
Copywriting and scripts
Build
Website mockups
UI design
Front-end development
Deployment
Workflow automation
MCP integrations
Operations
Inbox and Slack triage
Meeting-to-task capture
Weekday brief
End-of-day work sweep
Expense automation
Enablement
Role-based playbooks
AI training
Shared Claude Skills
Adoption surveys
Risk & governance
AI usage guidelines
Vendor due diligence
Certainty labels
Email and file forensics
Security and tool checks
The first 90 days
The five pillars above, put into sequence — in your organisation or a client’s. Each step has already been done once.
Days 1–15
Diagnose
Role-based baseline survey and 1:1 interviews; map where AI is already used, officially or not.
Done: 35 interviews analysed, survey cut from 18 to 5 questions.
Days 10–40
Govern
Usage policy, data rules and certainty labels agreed with legal, IT and security before scaling.
Done: Responsible AI guidelines and certainty labels in use.
Days 30–60
Prove
Three lighthouse use cases, each with a named owner and a before/after metric.
Done: RFQ response cut from 3–5 days to about 5 hours.
Days 45–75
Enable
Role-based playbooks, shared skills and a champion in every function.
Done: 12-lesson playbook, 13+ role guides, 5 shared skills.
Days 60–90
Measure & scale
Adoption and value reported to leadership; fund what works, stop what doesn’t.
Done: ~40 weekly active users; mandate approved with 30% time.