AI exposure in sales occupations
22 occupations, 427 tasks, scored against the capability frontier of 2026-09-15. This family ranks 3 of 22 by median exposed share.
Reading this family
computedThe median sales occupation has 49.7% of its weighted task load in work current AI systems can already produce, which is 25.4 points above the median across every occupation in the index. Weighted by how many people actually do these jobs, the figure is 49.4%, across 13,331,100 US workers. 21 of these occupations fall in the exposed band, 0 in assisted and 1 in untouched.
What holds the line across this family is context, at 1.67 out of 3, the highest of the five friction dimensions here. In plain terms the obstacle is knowing things a model cannot hold. The work runs on private, tacit or real-time organisational knowledge. This is the friction most likely to erode, because context is a plumbing problem rather than a law.
The spread inside the family is wider than the gap between families, which is the usual pattern and the reason this index does not publish a single number per group. Telemarketers sits at 73.3%. Models sits at 19.9%. Both are sales occupations, and 53.4 points separate them.
Where the family score comes from
judgedMeans across all 427 tasks in this family. Capability is what AI can do; the other five are what stands in the way.
| Dimension | Mean | Scale |
|---|---|---|
| Capability | 2.85 | 0-4 |
| Embodiment | 0.64 | 0-3 |
| Presence | 0.81 | 0-3 |
| Accountability | 0.81 | 0-3 |
| Context | 1.67 | 0-3 |
| Verification cost | 1.06 | 0-3 |
Every sales occupation
See them on the chartExposure is not displacement. These figures say what current systems can produce and what stands in the way. They do not say anyone will stop paying a person to do the work. How the score is built.