AI workforce planning
Only 19% of CHROs have built expected AI effects into enterprise-wide workforce and financial planning, and forecasting AI's effect on the workforce is their top barrier. The Conference Board, Q3 2026
AI workforce planning that starts from the tasks in each role.
AI changes the tasks inside a job long before it changes the job title. JobRoute scores every task in each of your roles for AI exposure from public data, counts the tasks scoring 60 or more, and rolls the result up into a role-by-role view your board can check.
Start with 10 job titles and no employee data. A founder sends a PDF in board-pack format within two business days.
One role, task by task
Accountants and Auditors
23 of 30 tasks score 60 or more
- Prepare, examine, or analyze accounting records, financial statements, or other financial... 40 Observed
- Process invoices for payment. 75 Estimated
- Review accounts for discrepancies and reconcile differences. 65 Estimated
- Prepare adjusting journal entries. 40 Observed
- Compute taxes owed and prepare tax returns, ensuring compliance with payment, reporting, or... 75 Estimated
Why plan from tasks
A whole-job view hides where change starts: only 73 of the 923 occupations with an O*NET task list reach an overall task exposure of 65, yet 790 of them, 85.6%, have at least one task that scores 60 or more.
Planning by task shows which parts of a role AI already does, so you can decide who to redeploy, retrain or retain before a title disappears from the org chart.
All 18,838 O*NET tasks, by AI exposure
- 60 or more: 5,321 tasks, 28.2%
- 41 to 59: 2,813, 14.9%
- 40 or less: 10,704, 56.8%
The method, from job title to board view
-
Match each job title to an occupation
Each title maps to one of the 1,016 occupations in O*NET, the U.S. Department of Labor's database of occupations and their tasks. Hard matches are flagged and confirmed with you.
-
Score every task from 0 to 100
Where the Anthropic Economic Index observes AI in use on a task, the score comes from that evidence. Where it does not, a documented rule based on the task's O*NET work activity sets the score, and the task records which method was used.
-
Count the tasks scoring 60 or more
Each role page counts the tasks that score 60 or more and says how many of those scores are observed in AI use. The role's exposure is the average across its tasks, weighted by how important O*NET rates each task, so the score leans toward the work people spend their days on.
-
Add the outlook for each role
Skill durability from the WEF Future of Jobs Report 2025, ten-year demand from BLS Employment Projections (2025 to 2035) and the nearby roles people could move into combine with exposure into the AI Ready Score, where higher means more AI-ready.
-
Roll up to people and cohorts
Your roster adds how many people hold each role. People with the same exposure pattern form cohorts, each with a recommended action: redeploy, retrain or retain. Each role also gets the AI model tier its tasks need.
-
Take it to the board
The board pack shows the workforce snapshot, the cohorts most affected and their recommended actions, the early-career pipeline gap and the reskilling business case, with a source for every number.
The full formula, weights and known limits are on the methodology page. Method v1.0.0, data version 2026-11-03.
The board view
A board needs one page that says, role by role, how much of the work AI already does and what the company will do about it. These rows are public data for common roles. In your board pack, each row also carries your headcount, the cohort and the recommended action.
| Role | Tasks at 60+ | Exposure | Model tier |
|---|---|---|---|
| Insurance Claims and Policy Processing Clerks | 23 of 25 | Standard | |
| Claims Adjusters, Examiners, and Investigators | 26 of 29 | Standard | |
| Accountants and Auditors | 23 of 30 | Standard | |
| Insurance Sales Agents | 11 of 19 | Frontier | |
| Software Developers | 10 of 17 | Frontier | |
| Actuaries | 12 of 15 | Frontier | |
| Human Resources Specialists | 15 of 26 | Frontier |
O*NET 31.0 tasks scored with the Anthropic Economic Index (June 2026). Model tier is the lowest tier that covers 80% of the work AI can address in the role. Data version 2026-11-03.
Start with the public data
Everything behind the method is public, so you can check it on roles you know before you send us anything.
- AI exposure by jobEvery task scored for 1,016 occupations, with BLS wages and outlook.
- AI exposure by stateThe share of workers in high-exposure occupations for all 50 states and DC.
- Data downloadsCSV and JSON, free with a verified email, reusable under CC BY 4.0.
- MethodologyThe formula, the four pillars, the sources and their releases.
- BLS AI exposure categoriesHow the August 2026 BLS ratings compare with task scores.
- The assessmentEvery role and person scored, cohorts and a board pack in about two weeks.
Questions about AI workforce planning
What is AI workforce planning?
It is workforce planning that accounts for the work AI can do. For each role, it asks which tasks AI already does, how many people do that work, and whether to redeploy, retrain or retain them, so headcount, hiring and training plans reflect how the work is changing.
How is task-based planning different from skills-based workforce planning?
Skills-based planning describes what people know. Task-based planning describes the work a role does, which is where AI shows up first: AI takes over tasks inside a job long before it removes a job title. JobRoute starts from the O*NET task list for each role and scores each task for AI exposure, then uses skills to judge where people can move.
What does a task scoring 60 or more mean?
Each O*NET task gets an AI exposure score from 0 to 100. At 60 or more, JobRoute counts the task as highly exposed to AI. Across all 1,016 occupations, 5,321 of 18,838 tasks score 60 or more. 1,145 task scores are observed in the Anthropic Economic Index (June 2026 release) and the rest are estimated from each task's O*NET work activity, and every job page shows which is which.
What data does JobRoute use for AI workforce planning?
Public sources: O*NET 31.0 for occupations and tasks, the Anthropic Economic Index (June 2026 release) for observed AI use by task, BLS Occupational Employment and Wage Statistics (May 2025) for employment and wages, BLS Employment Projections (2025 to 2035) for the outlook, and the WEF Future of Jobs Report 2025 for skill outlook. See the methodology.
Do you need our employee data?
Not to start. The free 10-role sample needs only job titles. The full assessment uses a roster export with job title, department and location, and no names, emails or salaries. Read how we handle your data.
How long does it take?
The free sample arrives within two business days. The full assessment takes about two weeks from the day we receive your roster and ends with a 60-minute board-prep session with a founder.
How often should an AI workforce plan be refreshed?
JobRoute re-runs the scores when O*NET, BLS or the Anthropic Economic Index publish a new release and records the new data version, and an optional annual subscription keeps a company's scores current.
Plan from the tasks, starting with 10 roles.
Send us up to 10 job titles, with no employee data. Within two business days you get a PDF in board-pack format and a 30-minute walkthrough with a founder.
A founder replies within one business day.