Bessemer Survey Finds AI Changing Work Before Headcount
Bessemer Venture Partners' May 18 survey of 173 leaders at 113 portfolio companies found 49% were delivering more without adding headcount and 13% had slowed or paused hiring. Claude was used by 73% of respondents, while adoption varied sharply by function, suggesting teams are changing workflows and staffing plans unevenly rather than following a single AI playbook.
Bessemer Venture Partners published findings on May 18 from a survey of 173 leaders across 113 of its portfolio companies, examining how six business functions are adopting artificial intelligence. The results show broad conviction about AI's impact, but much less consistency in how companies are putting it to work.
The survey found that 86% of leaders expected AI to meaningfully change how their teams operate over the next 12 months. Claude was the most-used tool across the portfolio, reported by 73% of respondents. Those figures describe Bessemer's portfolio sample, not the technology sector as a whole.
Engineering is ahead of other functions
Bessemer reported that 90% of technology and engineering teams either considered AI core to operations or were actively deploying it. Adoption was less mature elsewhere: 24% of finance teams were actively deploying AI, while 52% of people teams were still experimenting or getting started.
The obstacles also differed by function. Engineering leaders pointed to code quality, review bottlenecks, and difficulty measuring productivity. Finance teams emphasized fragmented systems and data quality, while people teams cited privacy and compliance. Marketing teams reported universal use of AI for content, but only 13% called it core to operations, with brand safety and quality control remaining prominent concerns.
Staffing effects are visible but uneven
Nearly half of respondents, 49%, said their teams were delivering more without adding headcount. Another 25% reported moving employees into AI-adjacent work, 13% had slowed or paused hiring, 10% had created new AI-workflow roles, and 6% had backfilled roles with AI tools. Bessemer also reported that 26% had seen no material workforce change.
These findings do not establish that AI caused every staffing decision. The study is a self-reported survey of venture-backed companies in one investor's portfolio, and the public page does not present it as a representative labor-market sample. Its practical value is narrower: it shows where early-adopting technology companies say AI is already changing workflows, which functions are moving fastest, and which operational barriers are slowing deployment.
For data and engineering leaders, the clearest signal is that tool availability is not the main dividing line. Teams are progressing differently because their data, review processes, compliance needs, and measurable workflows differ. Any benchmark drawn from the survey should therefore be compared function by function, not treated as one company-wide adoption score.
Key Points
- 1Bessemer surveyed 173 leaders across 113 portfolio companies and six business functions.
- 2Claude was used by 73% of respondents, while 90% of technology and engineering teams were deploying AI or considered it core.
- 3Forty-nine percent reported delivering more without adding headcount, but the portfolio survey is not a representative labor-market sample.
Scoring Rationale
First-party portfolio survey data provides useful function-level adoption and workforce benchmarks, with independent reporting confirming the published findings. The sample is limited to 173 leaders at 113 Bessemer portfolio companies, so the results are directional rather than representative of the wider labor market.
Sources
Primary source and supporting public references used for this report.
Practice with real Ad Tech data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Ad Tech problems