You have 1 article left to read this month before you need to register a free LeadDev.com account.
Estimated reading time: 3 minutes
Key takeaways:
- Set a baseline before you measure AI’s impact, using delivery metrics like DORA, then track what features generate once they ship.
- Fewer companies expect AI to shrink teams, with 55% planning to hire fewer developers in 2026, down from 76% in 2025.
- Bring your engineers into designing how AI is used, since 97% of tech leaders call human oversight important.
If you’re under pressure to show results from AI, you have company. Vention’s new survey of 305 technology leaders in the US and UK found that 96% of companies use AI in software development. Yet, only 55% have a clear strategy for it, and 57% feel pressure to implement AI even when the business case is unclear.
That last number won’t surprise anyone who has sat through a planning meeting this year. Plenty of teams adopted the tools first and are working out the justification now.
Your inbox, upgraded.
Receive weekly engineering insights to level up your leadership approach.
After AI adoption, the hard part is measurement
With AI adoption close to universal, attention has moved to what AI produces. Glyn Roberts, CTO of Digital Solutions at Vention, says the question clients bring him now is what to measure.
LeadDev’s AI Impact Report 2026 found the same gap. Only 31% of the 591 engineering leaders surveyed said their organization measures AI’s impact, and 68% named a lack of clear metrics as their biggest AI challenge.
Roberts advises starting with a baseline, using standard metrics like DORA, so you can see what actually changes. “Speed on its own proves nothing. If quality bars aren’t held, you’re just producing technical debt faster, and it soon becomes unmanageable.”
Delivery metrics only cover part of the picture. Roberts also points to business measures, such as how long a feature takes from idea to completion and what it generates once it’s live.
Fewer leaders expect AI to shrink their teams
The hiring numbers from Vention’s last four surveys trace an odd shape. In 2023, 66% of companies expected AI to let them hire fewer developers. That fell to about 60% in 2024, jumped to 76% in 2025 at the height of the coding assistant hype, then dropped to 55% in 2026.
So the tools got better, and confidence in cutting headcount went down. Slava Shahoika, head of AI practice at Vention, puts that down to how small a part of the work coding can be next to requirements analysis and solution design. AI “creates even higher demands on engineers as owners of the results,” Shahoika says.
That’s reassuring, with a catch: more than half of these companies still expect to hire fewer developers, and the cuts may land on juniors. In LeadDev’s AI Impact Report 2026, 62% of respondents expected fewer junior developers to be hired, up from 54% a year earlier.
More like this
Oversight is what keeps the speed
Nearly all of Vention’s respondents (97%) called human oversight an important part of AI-assisted development, and 75% called it very important. Oversight also came up as a trust factor, with 69% naming it as one of the main things that make AI-assisted development trustworthy.
Quality is climbing the list, too. The share of respondents who rate the quality of engineering execution as an important criterion for choosing a technology partner rose from 15% in 2025 to 23% in 2026.
Formal rules haven’t caught up. Only 46% of respondents in LeadDev’s AI Impact Report 2026 had adopted formal AI policy guidelines, yet 54% said their own responsibility for AI governance had grown, up from 34% in 2025.
Roberts argues that oversight only works when engineers help shape the process. “Accountability doesn’t move to the AI. The team is still responsible for what ships. That only works if engineers are involved in designing how AI is used, not just reviewing what it produces at the end.”
New York • September 8 & 9, 2027
Loved LDX3 New York? Pre-sale tickets for 2027 are now available.
For engineering managers, that’s a practical ask. Get your team into the room when AI norms are set, so the people reviewing the output also wrote the rules it has to meet.
The use cases have barely changed
Every year since Vention started its survey in 2023, productivity and quality have topped the list of reasons companies use AI.
In 2026, the leading use cases were automating routine tasks (49%), speeding up development (46%), and strengthening cybersecurity (42%).
After four years of new models and heavy investment, companies still value AI for making engineering work more efficient. What separates them now is whether they can show it.
