Your Accountant on the Scaffolding
Your accountant up on the scaffolding in January might sound like a fever dream — but that's essentially what our data is telling us. The Isle of Man's workforce resilience index tracks a metric called kappa: the ratio of AI adoption speed to the economy's ability to absorb displaced workers. When kappa exceeds 1.0, AI is displacing jobs faster than the island can redeploy people into new ones. Right now, kappa sits at 0.37 — manageable, but rising. The question isn't whether accountants will need to retrain. It's whether the scaffolding will be ready when they do.
Today we're launching Industry Deep Dives — comprehensive AI impact assessments for key Isle of Man industries, starting with Accountancy & Financial Professionals.
What Makes This Different
Most AI impact reports operate at 30,000 feet: "accountants face high automation risk." Helpful as a newspaper headline, useless as a policy input.
Our deep dives work at task level. We know which specific tasks an Isle of Man chartered accountant performs, which of those tasks current AI tools can automate, and — critically — which tools would do it. Not "AI might automate bookkeeping" but "Xero + Dext can automate receipt processing, bank reconciliation runs on automated matching rules, and monthly management accounts can be drafted by AI with human review."
This specificity comes from combining six data systems that don't normally talk to each other:
- Census 2021 — how many people actually work in these roles on the Isle of Man
- Live vacancy intelligence — what employers are hiring for right now, and what skills they want
- Anthropic's AI mode classification — automation vs augmentation share per occupation
- Frey-Osborne probabilities — the original Oxford automation risk scores, cross-walked to UK SOC codes
- Task-level AI/tool solutions — every job task classified as routine, augmented, or human-only, with specific AI tool recommendations for each
- Zapier's AI Fluency Framework — what "good" looks like at every level, from unacceptable to transformative
The Hiring Landscape Problem
One caveat we've surfaced that matters: many of the "employers" in our vacancy data are actually recruitment agencies advertising on behalf of undisclosed clients. The deep dive now flags these with an agency badge and adds a note about the data limitation. This means employer concentration metrics (HHI, top-3 share) reflect the agency market, not the actual employer market. It's an honest representation of what the data can and can't tell us.
AI Fluency: The Zapier Framework Applied
Each deep dive now includes an AI Fluency section built on Zapier's four-level framework. For every occupation in the sector, we show:
- Quick wins — things a professional can do this week to level up their AI practice
- Tools to learn — specific, named platforms relevant to their role
- A stand-out tip — the non-obvious thing that separates good from exceptional
- The fluency matrix — a task-by-task breakdown of what Unacceptable, Capable, Adoptive, and Transformative practice looks like
This isn't theoretical. A chartered accountant reading the matrix can immediately identify where they sit and what "one level up" looks like for their specific daily tasks.
Salary Data: What We Know and What We Don't
We've been transparent about data limitations. UK ONS ASHE salary data suppresses percentile values where sample sizes are too small — which means for many IoM-relevant SOC codes, only the median salary is reliable. The p10, p25, p75, and p90 columns show "suppressed" rather than misleading zeros.
This is an area where IoM-specific data collection would add significant value. The Government's own salary survey data, combined with our vacancy intelligence, could fill this gap entirely.
Stakeholder Recommendations
Every deep dive produces tailored recommendations for five audiences:
- UCM — curriculum changes, new programmes, industry partnerships
- Schools — career guidance, subject emphasis, work experience design
- Government — policy interventions, regulation, retraining investment
- Employers — AI adoption strategy, workforce planning, skills investment
- Workforce — personal upskilling, career pivots, tool mastery
These aren't generic. They reference the specific tools, salary data, and task compositions from the sector analysis. When the employer recommendation says "adopt automated bank reconciliation," it's because our data shows that specific task is classified as routine across 40% of live accountancy vacancies.
What's New on the Resilience Dashboard
Alongside the deep dives, the Workforce Resilience Index has had several updates based on stakeholder feedback:
- Income at Risk now uses labour income, not GDP — IoM GDP is ~70% corporate profits. Only ~30% (~GBP 2.3bn) flows to workers. Showing income at risk as a percentage of labour income rather than GDP gives a much more accurate picture of impact on actual workers.
- Vacancy data from our own scraper — the dashboard previously showed Government quarterly vacancy statistics. It now uses our own live scraped vacancy count, which updates daily and captures the full market.
- Sector transition pathways show real job titles — instead of abstract major group names, each transition pathway now lists example occupations on both sides, so you can see "Book-keepers, payroll managers" to "Business & financial project managers" rather than just SOC group labels.
- Economy-wide insights clarified — insights about kappa and transition character no longer show "43,308 affected workers" which was confusing since that's the entire workforce, not a specific at-risk subset.
Building the Next Deep Dive
The system is designed for repeatability. Each sector needs a configuration block (SOC codes, SIC codes, job categories), and the pipeline does the rest — census queries, AI exposure analysis, skills extraction, task-level mapping, salary data, employer landscape, and AI-generated assessment.
CSP (Corporate Service Providers) and Software Development are next. If your sector isn't listed yet, the deep dives page has a request section.
Data Sources
- IoM Census 2021 — workforce headcounts by SOC4 occupation
- Smart Island vacancy scraper — daily job listings with AI enrichment
- Anthropic Economic Index — task-level automation and augmentation scores
- Frey-Osborne / Oxford — automation probability per occupation
- FRS-AIOE — AI occupational exposure index
- ONS ASHE 2025 — UK salary benchmarks by SOC and SIC
- O*NET — occupation metadata, skills, knowledge, hot technologies
- Zapier AI Fluency Framework — four-level maturity model
Explore the Industry Deep Dives or start with the Accountancy assessment.