A Leanpreneur tool
£558.32bn of UK wages are exposed to AI
Every UK occupation has tasks an LLM can already do. This is a map of which ones, how exposed, and how much wage bill is in the firing line. Find your role, see the underlying tasks, then rebuild the work around the change rather than be replaced by it.
Scale note. The £558.32bn headline counts AI direct + AI with tools (γ reading). The stricter direct-replacement reading (α) is £107.2bn. The map below defaults to α, so judgment-heavy roles can look surprisingly low there even when AI helps with much of their workflow. You can flip the map to γ with the toggle above it.
direct-replacement (α) · other: 7.2/10
The map view is best on a wider screen. Here's the same data as a list — search or sort to find your role, then tap through.
299 roles
- Data entry administrators7.9/10
- Communication operators7.0/10
- Sales administrators6.9/10
- Other administrative occupations n.e.c.6.9/10
- Book-keepers, payroll managers and wages clerks6.5/10
- Project support officers6.4/10
- Data analysts6.4/10
- Business associate professionals n.e.c.6.4/10
- Health care practice managers6.0/10
- Medical secretaries6.0/10
- Telephone salespersons6.0/10
- Customer service occupations n.e.c.5.8/10
- Financial and accounting technicians5.7/10
- Financial accounts managers5.7/10
- Authors, writers and translators5.4/10
- Elementary administration occupations n.e.c.4.7/10
- Security system installers and repairers4.4/10
- Marketing and commercial managers4.3/10
- Advertising accounts managers and creative directors4.3/10
- Advertising and marketing associate professionals4.3/10
- Finance officers3.9/10
- Financial administrative occupations n.e.c.3.9/10
- Records clerks and assistants3.9/10
- Pensions and insurance clerks and assistants3.9/10
- Bus and coach drivers3.9/10
- Actuaries, economists and statisticians3.8/10
- Legal secretaries3.7/10
- Library clerks and assistants3.6/10
- Electronics engineers (professional)3.5/10
- Telecoms and related network installers and repairers3.5/10
- Newspaper, periodical and broadcast editors3.4/10
- Newspaper and periodical broadcast journalists and reporters3.4/10
- Transport and distribution clerks and assistants3.4/10
- National government administrative occupations3.3/10
- Local government administrative occupations3.3/10
- School secretaries3.3/10
- Personal assistants and other secretaries3.2/10
- Pharmaceutical technicians2.9/10
- Pharmacy and optical dispensing assistants2.9/10
- Train and tram drivers2.8/10
- Debt, rent and other cash collectors2.7/10
- Environmental health professionals2.6/10
- Health and safety managers and officers2.6/10
- Human resources administrative occupations2.6/10
- Sales related occupations n.e.c.2.6/10
- Routine inspectors and testers2.6/10
- Bank and post office clerks2.5/10
- Electrical engineers2.4/10
- Credit controllers2.4/10
- Business and related research professionals2.3/10
- Business, research and administrative professionals n.e.c.2.3/10
- Public relations professionals2.3/10
- Public services associate professionals2.3/10
- Receptionists2.2/10
- Youth and community workers2.1/10
- Child and early years officers2.1/10
- Housing officers2.1/10
- Welfare and housing associate professionals n.e.c.2.1/10
- Printers2.1/10
- Retail cashiers and check-out operators2.1/10
- Printing machine assistants2.1/10
- Leisure and sports managers and proprietors2.0/10
- Midwifery nurses2.0/10
- Prison service officers (below principal officer)2.0/10
- Production and process engineers1.9/10
- Higher education teaching professionals1.9/10
- Special and additional needs education teaching professionals1.9/10
- Electrical and electronics technicians1.9/10
- Stock control clerks and assistants1.9/10
- Engineering project managers and project engineers1.8/10
- Engineering professionals n.e.c.1.8/10
- Quantity surveyors1.8/10
- Clergy1.8/10
- Quality control and planning engineers1.8/10
- Call and contact centre occupations1.8/10
- Assemblers (vehicles and metal goods)1.8/10
- Assemblers and routine operatives n.e.c.1.8/10
- Heavy and large goods vehicle drivers1.8/10
- Postal workers, mail sorters and messengers1.8/10
- Education advisers and school inspectors1.7/10
Showing the top 80 — refine the search to find a specific role.
Each tile is a UK occupation. Area shows the total wage bill (employees × median pay). Colour shows AI exposure on a 0–10 scale. The biggest 60 occupations are named; smaller ones within each major group are collapsed into an "Other [group]" tile to keep the map legible. Click any named tile to see the underlying tasks.
Why do solicitors, CEOs, and marketing directors look low-exposure?
Exposure is task-level. Senior roles are dominated by decision-making, relationship work, and judgment calls - the things Eloundou's annotators correctly label "not direct LLM replacement". But those same roles do a lot of drafting, analysis and review where AI helps massively. The α reading is strict; the γ reading (shown in the stats row and on each role page) captures the with-tools picture. For solicitors that's α=0.0 / γ=8.5 - near the top of the range when you include AI with the firm's own documents, templates and search.
One honest caveat: the underlying task labels were made in 2023 against early GPT-4. Frontier models can do materially more in 2026, so even γ is probably a floor on what's possible today. A fresh Claude-scored re-rating is on the roadmap (methodology).
How to read it
Each tile is an occupation. Bigger tile = more employees × pay. Darker colour = higher exposure to AI on a 0–10 scale. Click a tile for the underlying tasks.
Scores come from Eloundou et al (2023), who labelled ~19,000 O*NET work tasks by how much a language model can do them. The tile colours use the strict "direct replacement" reading (0–10 per occupation). The big headline at the top uses a more generous reading that includes tasks an LLM can do when paired with tools, which is closer to how AI is actually being used in 2026.
Employment and pay come from ONS ASHE 2025. That covers employees only - self- employed workers aren't in the wage bill, so freelance-heavy trades are under-weighted.
Read it as a map, not a verdict. A high score doesn't mean the role disappears. It means a lot of the task inventory is touchable by AI, and that the useful question is "which of my tasks can I stop doing, and what do I do with the time." That's what the drilldowns and the community are for.
Top 10 UK occupations by wage exposure
Wage exposure equals exposure score × employees × median gross annual pay. Not "roles that will disappear" - these are the places where the most money meets the most exposure.
- Other administrative occupations n.e.c.†
- Book-keepers, payroll managers and wages clerks
- Customer service occupations n.e.c.†
- Financial accounts managers
- Marketing and commercial managers
- Higher education teaching professionals
- Business associate professionals n.e.c.†
- Sales and retail assistants
- Production managers and directors in manufacturing
- Advertising and marketing associate professionals
† Residual SOC 2020 categories - "not elsewhere classified". They aggregate smaller occupations that don't fit the named unit groups, so a high score is an average across a mixed bag of task types. Treat with appropriate suspicion.
Sources and methodology
- ONS SOC 2020 Volume 1 — structure and descriptions of unit groups · 2025-12-03 · OGL v3.0
- ONS SOC 2020 Volume 2 — coding index (contains SOC 2020 ↔ ISCO-08) · 2025-12-03 · OGL v3.0
- O*NET 30.2 database (tab-delimited, ZIP) · 30.2 · CC BY 4.0
- Eloundou et al — per-task AI exposure scores (full_labelset.tsv, alpha = direct exposure) · 0471612f (pinned) · MIT (repo)
- US BLS — ISCO-08 ↔ US SOC 2010 crosswalk · 2012 (crosswalk current) · Public Domain (US Government work)
- ONS ASHE Table 14 — occupation (4-digit SOC 2020), 2025 provisional (includes employment count and gross annual pay) · 2025 provisional (released 2025-10-23) · OGL v3.0
Built 23 April 2026. Coverage 91.99% of UK SOC 2020 unit groups (379 of 412).