AI and Jobs: What Data Shows for Sales Tech
By Stanislav Chirk3 min read
Headlines say AI will take your job. The data says something else.
Anthropic's March 2026 labor market study introduces observed exposure: not what AI could do, but what it actually does in professional settings. The difference matters for anyone thinking about sales automation.
The gap: theory vs reality
94% → 33%
Computer & Math: theoretical vs observed AI coverage
Anthropic (2026)
AI covers a fraction of what it theoretically could. In Computer & Math occupations, theory suggests 94% of tasks are LLM-feasible — real observed coverage is 33%.
The study combines:
- O*NET task data (US occupations)
- Anthropic Economic Index (actual Claude usage)
- Eloundou et al. (2023) theoretical exposure scores
Result: a measure that weights automated use more than augmentative, and work-related context over casual use. Observed exposure is lower than you'd expect from capability hype.
Source: Anthropic, Labor market impacts of AI, Figure 2. Reproduced under fair use.
Who's exposed, who isn't
High exposure (examples): Computer programmers, customer service reps, data entry keyers, financial analysts. Common thread: structured, repetitive, language-heavy tasks that show up in real AI traffic.
Low or zero exposure (~30% of workers): Cooks, mechanics, bartenders, lifeguards. Physical, situational, or highly contextual work.
Sales-adjacent roles sit in the middle. Qualification, first-line configuration, routine proposal drafting — exposed. Relationship selling, negotiation, complex exceptions — less so.
No unemployment spike (yet)
The study finds no systematic increase in unemployment for highly exposed workers since late 2022. There is, however, suggestive evidence that hiring of younger workers (22–25) into exposed occupations has slowed.
Interpretation: displacement isn't showing up as layoffs. It may show up as fewer new hires. Early signal, not conclusive.
What this means for sales tech
Presales and qualification are classic "structured but contextual" tasks. Perfect for AI augmentation, not full replacement:
- Co-Sales.AI Configurator — interviews the prospect, validates against the catalog, delivers a quote. The agent handles the structured part; the human handles edge cases, relationship, and sign-off.
- Talkulate — turns a short conversation into a personalized proposal. The AI collects context and runs your logic; the rep owns the relationship and closing.
The gap between theory and observed usage is where most sales automation lives today. Tools that augment clear workflows (config, quote, qualification) thrive. The human remains in the loop for judgment and trust.
Bottom line
- AI's observed impact lags behind its theoretical capability.
- High-exposure jobs haven't seen a visible unemployment spike so far.
- Sales roles: automate the structured work (config, proposal, qualification); keep humans for relationship and exceptions.
That's what the data shows. The hype is ahead of reality — which leaves room for tools that get the balance right.