Explore what they hand over to AI and where they draw the line.
How we measure it
The Delegation Ladder
Six levels of AI delegation. From doing things yourself to letting AI own the task.
Code review at each level of AI involvement
L0Not used0% of work is delegated to AI
L1Tool
L2Assistant
L3General-purpose Executor
L4Supervised Executor
L5Competence Center
LEVEL 0
Manual work
You review the code manually.
You: Doer
You perform the entire task manually.
AI: Not involved
LEVEL 1
On-Demand Assistance
AI explains diffs or flags style issues upon request.
You: Operator
You drive the process and use AI
for specific, isolated queries.
AI: Tool
It provides specific answers or actions only when triggered.
LEVEL 2
AI as a drafter
AI drafts comments for specific files. You review and post them.
You: Collaborator
You define the tasks and compile
the final output from the AI drafts.
AI: Assistant
It generates drafts or suggestions
for these tasks.
LEVEL 3
Delegated execution
AI reviews the entire PR and proposes comments. You approve them.
You: Consultant
You review and approve the complete work produced by the AI.
AI: General-purpose Executor
It executes the complete task and presents results for review.
LEVEL 4
Supervised Autonomy
AI automatically reviews PRs. You only intervene for critical flags.
You: Approver
You set the direction and approve key milestones.
AI: Supervised Executor
It works autonomously and surfaces only critical decisions.
LEVEL 5
Full AI Ownership
AI handles all code reviews based on your defined standards.
You: Stakeholder/Auditor
You define the standards and audit the results.
AI: Competence Center
It owns the entire process end‑to‑end.
Dive into the data
Different lenses on the AI delegation ladder
The same ladder, different ways to analyze it. Pick a lens to open the full view.
LENS · 01 / 05
Map
Where each tool sits on the ladder, by activity.
See the spreadEach tool's users across the five levels.
Find the consensusWhere the market agrees on trust.
ExportDownload as CSV or PNG.
“
The last two years were pivotal in shaping the decision of developers to use AI. The next two define how much of their work they will hand over. This transition, not model benchmarks, will determine the market.”
Market Intelligence Team· JetBrains
Credits
Kairos is made by JetBrains SRMI — Strategic Research & Market Intelligence — led by Mariia Antropova. This work comes out of the AI Domain (delegation and the AI Pulse survey) and Technology & Innovation Insights (trends).
Mariia Antropova
Head of SRMI
Stanislav Tikhomirov
AI Domain Lead
Constantin Kichinsky
Technology & Innovation Insights
Misha Bogdanov
Quant Surveys & Research
Sasha Podborodnikova
Data Science
Karina Subkhangulova
UX Research
Darya Andreeva
UX Research
Innokentiy Burtsev
UX Research
Sofia Kulikova
Marketing Research
Olga Lvova
DX and Productivity Research
Don't miss the shift
The shifts that reach your roadmap six months from now
New evidence, revised forecasts, and one study worth your time. For anyone building with AI. Free. Unsubscribe anytime