SlashData

AI Analysts & Developer Research
We look at the data to uncover trends and insights, so Tech leaders can navigate tech decisions with confidence & clarity.

[SlashData @SlashDataHQ]

Cloud native developers reached 21.2 million, about 41% of all developers (19.9 million in Q1 2026). 67% add AI to their services, against 38% of others. 32% of both groups run agents in production (n=7,782). State of Cloud Native Development, for @CNCF: https://t.co/02Ht8HgCoB

[SlashData @SlashDataHQ]

An AI that writes CI/CD pipelines sounds dull. For a developer with a busy workflow, it can be "a huge burden off their hands."

Liam Bollmann-Dodd on why developers rely on AI most for the tasks they would rather not do:
https://t.co/IqHL6ONcQ1

[SlashData @SlashDataHQ]

With AI agents, the hard part of cost may be predicting it. A task "could be two minutes, it could be an hour and a half."

Liam Bollmann-Dodd's reading, which he is clear is inference, with nothing specific in the data:
https://t.co/Lyj3dW4OaX

[SlashData @SlashDataHQ]

AI is jagged: in what it can do, and in how useful it is to each developer.

Liam Bollmann-Dodd on why the same AI feature can matter a lot to one developer and little to the next:
https://t.co/FKGLj8HtmI

[SlashData @SlashDataHQ]

9 to 12 November 2026: we will be at Web Summit in Lisbon and at KubeCon + CloudNativeCon North America in Salt Lake City. Book a meeting in Lisbon and get our premium Future of Development Teams report.

Book a meeting:
https://t.co/AMB5xvLnB1

#WebSummit #KubeCon #SlashData

[SlashData @SlashDataHQ]

Fewer developers hand code comments to AI as a separate task: 34% to 26%.

Base: developers who use AI tools to write code, global survey of 10,000+ developers per wave, Q1 vs Q3 2026.

Liam Bollmann-Dodd's reading: AI writes them as it writes the code.
https://t.co/JfpcNrmphr

[SlashData @SlashDataHQ]

Reliance on AI to find security vulnerabilities rose from 1 to 8 in six months.

Base: developers who use AI tools to write code, global survey of 10,000+ developers per wave, Q1 vs Q3 2026.

Liam Bollmann-Dodd on why:
https://t.co/xvI0pUwD6d

[SlashData @SlashDataHQ]

The new https://t.co/hSUwMwDjbM is live. Here is what you can do on it:

• Read our free reports, each with the full analysis and the sample behind it
• Talk to an analyst about the question you need answered

Explore the site:
https://t.co/E618vpwZ7K

[SlashData @SlashDataHQ]

Reliance on AI for DevOps tasks swung from -9 to +9. The share using AI for them barely moved: 19% to 17%.

Base: developers who use AI tools to write code, global survey of 10,000+ developers per wave, Q1 vs Q3 2026.

Liam Bollmann-Dodd on why:
https://t.co/PVgFMYk0uY

[SlashData @SlashDataHQ]

Code generation is no longer the top AI task: 49% to 41%. Those still using it rely on AI more: -8 to -1.

Base: developers who use AI tools to write code, global survey of 10,000+ developers per wave, Q1 vs Q3 2026.

Liam Bollmann-Dodd explains why:
https://t.co/tn3TlzYEQ5

[SlashData @SlashDataHQ]

@MrrrOzi This is true!

[SlashData @SlashDataHQ]

62% of developers with 16+ years of experience say AI developer tools cut their time on repetitive tasks. With 0-2 years, it is 42% less time against 44% more.

Base: 4,203 professional developers, Developer Nation Q1 2026.

https://t.co/rxJDpjZUv4

[SlashData @SlashDataHQ]

Great question. We can't isolate the "already in the IDE" effect directly, but there are strong signals. Copilot has the highest share of users whose organisation mandates it and the highest company-funded share. And yes, satisfaction tells a different story: Claude Code and Codex lead on CSAT over Copilot.

[SlashData @SlashDataHQ]

The AI Developer Tools Benchmark Q3 2026 is out! See how 19 AI developer tools compare on adoption, conversion, and satisfaction.

GitHub Copilot converts 54% of aware developers; Codex and Claude Code 44%. Base: 2,063 professional developers, Q3 2026.

https://t.co/nuX7kV63ch

[SlashData @SlashDataHQ]

34% of developers say difficulty estimating cost limits how much autonomy they give AI agents.

Agents are being metered. The advantage goes to the vendor whose meter is easiest to read.

https://t.co/Hdemkqbq5h
Source: SlashData AI Developer Tools Benchmark, Q3 2026 (n=2,040)

[SlashData @SlashDataHQ]

Tomorrow, 16:00 BST: the AI tasks developers are quietly dropping.

About to roll out a new AI tool or AI feature? See which tasks to hand to AI first. Liam Bollmann-Dodd and Petro Ylli, 45 minutes, questions included. The room holds 100.

https://t.co/1tiXPnoTmH

[SlashData @SlashDataHQ]

Only 14% of developers let an AI agent run end to end with minimal or no review.

Bounded autonomy is the most common way developers use agentic tools, so agent messaging lives between those two points. Base: professional developers, Q3 2026 wave.

https://t.co/iXQCEK57Ky

[SlashData @SlashDataHQ]

GitHub Copilot converts 48% of the developers who know it into users. Most tools convert fewer than one in five.

The AI developer tools in the Q3 awareness-to-usage funnel, base professional developers, Q3 2026 wave. Full wave out in early October.

https://t.co/VkaUIGePY0

[SlashData @SlashDataHQ]

Know how developers use AI before you build your next AI tool.

Developers are giving AI fewer tasks, but relying on it more for the ones they keep. Liam Bollmann-Dodd shows which ones, live. The room holds 100.

Thu 8 Oct, 16:00 to 16:45 BST.

https://t.co/57WEzzqiU8

[SlashData @SlashDataHQ]

One category label, five very different customers.

A user on Wix Harmony, Lovable, Replit, Claude Code or OpenAI Codex is not the same type of customer.

Collapsing them into one category distorts TAM, positioning, onboarding, pricing and governance.

https://t.co/BtcvWAKUd6

#AIBuilders #ProductStrategy