AI usage patterns in software teams
From aiste.ulozaite@gmail.com · original ↗ · unsubscribe
Tens of thousands of teams build software inside Linear every day. The company is unusually well placed to see the entire workflow behind building a product. This post provides a picture of AI adoption within Linear’s customer base. It looks at who is using AI, how it reshapes where teams spend their time across Linear, and whether it has changed how much they ship.
How teams build – Linear
HOW TEAMS BUILD
AI usage patterns in software teams
0000000000000000000000000000000000000000000000000111111111111111111111111111111111111111111111111122222222222222222222222222222222222222222222222223333333333333333333333333333333333333333333333444444444444444444444444444444444444444444444444555555555555555555555555555555555555555555555555566666666666666666666666666666666666666666666667777777777777777777777777777777777777777777777778888888888888888888888888888888888888888888888899999999999999999999999999999999999999999999999992026Linear Orbit Inc.
EDITION 01 - TIM QI (2026)
Tens of thousands of teams build software inside Linear every day. Over six years that’s given us a detailed picture of how product development happens, from before AI was widely adopted to now.
Model companies and coding tools have published plenty on token usage and code volume, but that captures only one layer of the work. We’re unusually well placed to see the entire workflow behind building a product, from the first issue to the pull request that closes it. What we can’t see is AI usage that happens outside Linear, so this is a picture of adoption inside our own customer base, not the market at large.
We look at three things across that transition. Who is using AI, how it reshapes where teams spend their time across Linear, and whether it changes how much they ship. Together they make a fixed point for where AI-assisted product development stands in 2026, and something to measure the next edition against.
Adoption
Application
Output
Adoption by function
AI adoption has spread to every function
Between January and June 2026 the share of users active on AI features more than doubled in every function. Product climbed fastest, from 12% to 34%, and even go-to-market, the function furthest from the codebase, went from 5% to 18%. We classify roles by normalizing job titles, which carries some error at the edges, but the pattern is too broad to be an artifact of labeling.
Percentage of users active on Linear AI features (Last 30 days) by function
Founder
+16pp
Engineering
+18pp
Product
+22pp
Design
+16pp
GTM
+13pp
0%10%20%30%40%50%
Percentage of users active on Linear AI features in the last 30 days, by function
Segment
Jan 2026
Jun 2026
Change
Founder
14%
30%
+16 percentage points
Engineering
12%
30%
+18 percentage points
Product
12%
34%
+22 percentage points
Design
6%
22%
+16 percentage points
GTM
5%
18%
+13 percentage points
Jan 2026Jun 2026More infoN = 127,000 paid users, active in both January and June 2026
Adoption by executive team
Adoption goes all the way to the top
Executives are personally active on AI at rates that match or beat their teams. CEOs at companies of 201 or more people went from 9% to 36% in six months, the largest jump of any cut in this report, suggesting the most senior leaders are learning the technology by using it rather than reading about it. Company size comes from third-party enrichment, so this cut covers fewer workspaces than the rest of the report.
Percentage of users active on Linear AI features (Last 30 days) by executive team
Founder201+
+16pp
51-200
+12pp
1-50
+16pp
CEO201+
+27pp
51-200
+11pp
1-50
+14pp
CPO201+
+21pp
51-200
+15pp
1-50
+25pp
CTO201+
+24pp
51-200
+16pp
1-50
+17pp
0%10%20%30%40%50%
Percentage of users active on Linear AI features in the last 30 days, by executive team
Segment
Jan 2026
Jun 2026
Change
Founder, 201+
10%
26%
+16 percentage points
Founder, 51-200
15%
27%
+12 percentage points
Founder, 1-50
15%
31%
+16 percentage points
CEO, 201+
9%
36%
+27 percentage points
CEO, 51-200
15%
25%
+11 percentage points
CEO, 1-50
7%
21%
+14 percentage points
CPO, 201+
3%
24%
+21 percentage points
CPO, 51-200
10%
26%
+15 percentage points
CPO, 1-50
11%
36%
+25 percentage points
CTO, 201+
11%
35%
+24 percentage points
CTO, 51-200
12%
28%
+16 percentage points
CTO, 1-50
16%
33%
+17 percentage points
Jan 2026Jun 2026More infoN = 13,300 executives, active in both January and June 2026
Adoption by company size
Adoption is consistent at every size
AI adoption roughly tripled everywhere, from startups to enterprises. Company size, usually a good predictor of how fast an organization moves on new technology, barely registers here.
Percentage of users active on Linear AI features (Last 30 days) by company size (employees)
1001+ FTE
+17pp
201-1000 FTE
+19pp
51-200 FTE
+16pp
1-50 FTE
+14pp
0%10%20%30%40%50%
Percentage of users active on Linear AI features in the last 30 days, by company size in full-time employees
Segment
Jan 2026
Jun 2026
Change
1001+ FTE
8%
25%
+17 percentage points
201-1000 FTE
9%
27%
+19 percentage points
51-200 FTE
9%
25%
+16 percentage points
1-50 FTE
8%
23%
+14 percentage points
Jan 2026Jun 2026More infoN = 199,000 paid users with a known company size, active in both January and June 2026
Application - Create & organize
Teams are putting more into the system
Between June 2025 and June 2026, time spent creating, triaging, and commenting rose in nearly every function, with engineering up roughly 17% on create and triage alone. Founders show much larger swings, up 17 minutes on creation and 26 on commenting, though they’re a smaller cohort and noisier for it. More work seems to need more coordination, and that coordination increasingly sets the context agents act on.
Average minutes per user per month, June 2025 vs June 2026
Create & triageEng
+5m
Product
-1m
Design
+3m
GTM
+4m
Founder
+17m
Assign & updateEng
+3m
Product
0m
Design
+3m
GTM
+3m
Founder
+7m
CommentEng
+5m
Product
+1m
Design
+2m
GTM
+6m
Founder
+26m
0m15m30m45m60m75m
Average minutes spent per user creating, triaging, assigning, updating, and commenting on issues, June 2025 versus June 2026, by function
Segment
Jun 2025
Jun 2026
Change
Create & triage, Eng
24m
28m
+5 minutes
Create & triage, Product
38m
37m
-1 minutes
Create & triage, Design
22m
25m
+3 minutes
Create & triage, GTM
27m
31m
+4 minutes
Create & triage, Founder
40m
57m
+17 minutes
Assign & update, Eng
16m
19m
+3 minutes
Assign & update, Product
26m
26m
0 minutes
Assign & update, Design
12m
15m
+3 minutes
Assign & update, GTM
12m
15m
+3 minutes
Assign & update, Founder
22m
29m
+7 minutes
Comment, Eng
35m
40m
+5 minutes
Comment, Product
48m
49m
+1 minutes
Comment, Design
32m
34m
+2 minutes
Comment, GTM
49m
55m
+6 minutes
Comment, Founder
39m
64m
+26 minutes
Jun 2025Jun 2026More infoN = 54,300 paid users (Jun 2025) → 89,000 (Jun 2026)
Application - Issue creation
AI authors nearly half of all issues
Two years ago, fewer than one issue in a thousand was created by AI. Teams now use AI to write just under half of everything created in Linear, and at the current pace it will soon author more than people and integrations combined.
Issues created per week (thousands) by source
300025002000150010005000
Jul 2024Jan 2025Jul 2025Jan 2026Jul 2026
Thousands of issues created per week by agents and MCP clients versus people and integrations, June 2024 to August 2026, excluding imported issues
Week of
Agents & MCP
People & integrations
Jun 3, 2024
0
605
Jun 10, 2024
0
599
Jun 17, 2024
0
582
Jun 24, 2024
0
689
Jul 1, 2024
0
602
Jul 8, 2024
0
628
Jul 15, 2024
1
621
Jul 22, 2024
0
627
Jul 29, 2024
1
650
Aug 5, 2024
0
654
Aug 12, 2024
1
624
Aug 19, 2024
0
660
Aug 26, 2024
1
650
Sep 2, 2024
1
670
Sep 9, 2024
1
690
Sep 16, 2024
1
692
Sep 23, 2024
1
725
Sep 30, 2024
0
696
Oct 7, 2024
1
726
Oct 14, 2024
1
724
Oct 21, 2024
1
741
Oct 28, 2024
1
721
Nov 4, 2024
1
760
Nov 11, 2024
0
760
Nov 18, 2024
1
795
Nov 25, 2024
1
677
Dec 2, 2024
1
765
Dec 9, 2024
1
800
Dec 16, 2024
1
770
Dec 23, 2024
0
373
Dec 30, 2024
0
460
Jan 6, 2025
1
825
Jan 13, 2025
1
878
Jan 20, 2025
1
869
Jan 27, 2025
1
920
Feb 3, 2025
1
930
Feb 10, 2025
1
924
Feb 17, 2025
1
890
Feb 24, 2025
1
934
Mar 3, 2025
1
942
Mar 10, 2025
1
974
Mar 17, 2025
3
971
Mar 24, 2025
3
984
Mar 31, 2025
1
985
Apr 7, 2025
1
999
Apr 14, 2025
1
974
Apr 21, 2025
1
994
Apr 28, 2025
3
1029
May 5, 2025
5
1037
May 12, 2025
7
1063
May 19, 2025
9
1042
May 26, 2025
11
992
Jun 2, 2025
18
1095
Jun 9, 2025
18
1074
Jun 16, 2025
28
1064
Jun 23, 2025
34
1148
Jun 30, 2025
35
1092
Jul 7, 2025
44
1166
Jul 14, 2025
40
1137
Jul 21, 2025
41
1164
Jul 28, 2025
45
1177
Aug 4, 2025
52
1171
Aug 11, 2025
50
1206
Aug 18, 2025
55
1177
Aug 25, 2025
47
1225
Sep 1, 2025
48
1200
Sep 8, 2025
46
1299
Sep 15, 2025
45
1272
Sep 22, 2025
45
1294
Sep 29, 2025
58
1338
Oct 6, 2025
62
1352
Oct 13, 2025
68
1350
Oct 20, 2025
65
1382
Oct 27, 2025
74
1414
Nov 3, 2025
85
1461
Nov 10, 2025
85
1457
Nov 17, 2025
91
1436
Nov 24, 2025
93
1299
Dec 1, 2025
122
1473
Dec 8, 2025
142
1487
Dec 15, 2025
147
1526
Dec 22, 2025
110
795
Dec 29, 2025
139
808
Jan 5, 2026
206
1601
Jan 12, 2026
273
1725
Jan 19, 2026
291
1721
Jan 26, 2026
323
1806
Feb 2, 2026
401
1897
Feb 9, 2026
451
1901
Feb 16, 2026
516
1875
Feb 23, 2026
599
2048
Mar 2, 2026
707
2123
Mar 9, 2026
794
2170
Mar 16, 2026
837
2106
Mar 23, 2026
916
2297
Mar 30, 2026
935
2104
Apr 6, 2026
1038
2063
Apr 13, 2026
1128
2297
Apr 20, 2026
1209
2173
Apr 27, 2026
1275
2185
May 4, 2026
1382
2238
May 11, 2026
1506
2278
May 18, 2026
1597
2271
May 25, 2026
1472
2132
Jun 1, 2026
1542
2270
Jun 8, 2026
1766
2371
Jun 15, 2026
1652
2256
Jun 22, 2026
1728
2372
Jun 29, 2026
1799
2265
Jul 6, 2026
2078
2532
Jul 13, 2026
2143
2465
Jul 20, 2026
2195
2396
Jul 27, 2026
2348
2357
Aug 3, 2026
2435
2481
Agents & MCPPeople & integrationsMore infoIssues created per week, June 2024 to August 2026. Excludes imported issues
Application - Planning
Planning time didn’t move inside Linear
Time spent on customer requests, docs, and projects held steady in a year when nearly everything else in this report moved up. Planning practice varies widely from team to team, and plenty of it happens in conversation before it lands anywhere, so the average blends heavy planners with light ones. What the steadiness suggests is that AI has so far changed how teams execute far more than how they decide what to build.
Average minutes per user per month, June 2025 vs June 2026
Customer requestsEng
0m
Product
0m
Design
0m
GTM
+1m
Founder
+1m
Docs & projectsEng
+1m
Product
+1m
Design
+1m
GTM
+1m
Founder
0m
0m5m10m15m20m25m
Average minutes spent per user on customer requests, docs, and projects, June 2025 versus June 2026, by function
Segment
Jun 2025
Jun 2026
Change
Customer requests, Eng
1m
1m
0 minutes
Customer requests, Product
3m
4m
0 minutes
Customer requests, Design
1m
1m
0 minutes
Customer requests, GTM
4m
4m
+1 minutes
Customer requests, Founder
2m
3m
+1 minutes
Docs & projects, Eng
3m
3m
+1 minutes
Docs & projects, Product
13m
14m
+1 minutes
Docs & projects, Design
4m
5m
+1 minutes
Docs & projects, GTM
3m
3m
+1 minutes
Docs & projects, Founder
7m
8m
0 minutes
Jun 2025Jun 2026More infoN = 54,300 paid users (Jun 2025) → 89,000 (Jun 2026)
Application - AI
A new layer of work appeared
Chatting with AI and delegating issues to agents are categories of work that didn’t exist a year ago, and they now show up in every function’s week, with product leaning in hardest. Nothing else shrank to make room, which suggests AI has landed on top of existing work rather than replacing any of it, at least so far.
Average minutes per user per month, June 2025 vs June 2026
Agent issuesEng
+1m
Product
+1m
Design
0m
GTM
0m
Founder
+2m
Chat with AIEng
+2m
Product
+5m
Design
+3m
GTM
+3m
Founder
+4m
0m2m4m6m8m10m
Average minutes spent per user on agent issues and AI chat, June 2025 versus June 2026, by function
Segment
Jun 2025
Jun 2026
Change
Agent issues, Eng
0m
1m
+1 minutes
Agent issues, Product
0m
1m
+1 minutes
Agent issues, Design
0m
0m
0 minutes
Agent issues, GTM
0m
0m
0 minutes
Agent issues, Founder
0m
2m
+2 minutes
Chat with AI, Eng
0m
2m
+2 minutes
Chat with AI, Product
0m
5m
+5 minutes
Chat with AI, Design
0m
3m
+3 minutes
Chat with AI, GTM
0m
3m
+3 minutes
Chat with AI, Founder
0m
4m
+4 minutes
Jun 2025Jun 2026More infoN = 54,300 paid users (Jun 2025) → 89,000 (Jun 2026)
Output - PR creation
Non-engineers are shipping more code
The share of product managers attaching pull requests rose from 3% to 10% in two years, and designers from 1% to 8%. We only count pull requests in repositories connected to Linear, so anyone shipping outside that loop is invisible here, which makes these numbers floors rather than ceilings. The people who used to describe a change increasingly ship it themselves.
Percentage of users who attached a pull request (Last 30 days)
Founder
+12pp
Engineering
+14pp
Product
+7pp
Design
+7pp
GTM
+2pp
0%10%20%30%40%50%
Percentage of users who attached a pull request in the last 30 days, June 2024 to June 2026, by function
Segment
Jun 2024
Jun 2025
Jun 2026
Change
Founder
11%
12%
23%
+12 percentage points
Engineering
20%
22%
34%
+14 percentage points
Product
3%
3%
10%
+7 percentage points
Design
1%
2%
8%
+7 percentage points
GTM
1%
1%
3%
+2 percentage points
Jun 2024Jun 2025Jun 2026More infoN = 166,000 paid users (June 2026)
Output - PR volume
Pull requests are up 111% in two years
Pull requests opened per workspace are up 111% on a June 2024 baseline. Output held roughly level for the first year, then bent upward through 2026 as model quality and adoption climbed together. We count PRs opened rather than merged, and an opened PR says nothing about the value of the change, but the inflection is hard to miss.
Percentage change in pull requests per team per week since June 2024 - All paid workspaces
+125%+100%+75%+50%+25%0%-25%-50%-75%
Jul 2024Jan 2025Jul 2025Jan 2026Jul 2026
Weekly change in pull requests opened per paid workspace, against the June 2024 baseline
Week of
Change
Jun 2, 2024
0%
Jun 9, 2024
+9%
Jun 16, 2024
+10%
Jun 23, 2024
+3%
Jun 30, 2024
+8%
Jul 7, 2024
-4%
Jul 14, 2024
+8%
Jul 21, 2024
+7%
Jul 28, 2024
+8%
Aug 4, 2024
+7%
Aug 11, 2024
+6%
Aug 18, 2024
+3%
Aug 25, 2024
+10%
Sep 1, 2024
+10%
Sep 8, 2024
+5%
Sep 15, 2024
+12%
Sep 22, 2024
+10%
Sep 29, 2024
+14%
Oct 6, 2024
+8%
Oct 13, 2024
+11%
Oct 20, 2024
+9%
Oct 27, 2024
+18%
Nov 3, 2024
+8%
Nov 10, 2024
+15%
Nov 17, 2024
+11%
Nov 24, 2024
+17%
Dec 1, 2024
0%
Dec 8, 2024
+16%
Dec 15, 2024
+17%
Dec 22, 2024
+10%
Dec 29, 2024
-58%
Jan 5, 2025
-50%
Jan 12, 2025
+6%
Jan 19, 2025
+15%
Jan 26, 2025
+13%
Feb 2, 2025
+15%
Feb 9, 2025
+19%
Feb 16, 2025
+21%
Feb 23, 2025
+17%
Mar 2, 2025
+21%
Mar 9, 2025
+19%
Mar 16, 2025
+26%
Mar 23, 2025
+26%
Mar 30, 2025
+23%
Apr 6, 2025
+17%
Apr 13, 2025
+24%
Apr 20, 2025
+11%
Apr 27, 2025
+10%
May 4, 2025
+7%
May 11, 2025
+14%
May 18, 2025
+21%
May 25, 2025
+22%
Jun 1, 2025
+9%
Jun 8, 2025
+22%
Jun 15, 2025
+16%
Jun 22, 2025
+12%
Jun 29, 2025
+22%
Jul 6, 2025
+8%
Jul 13, 2025
+16%
Jul 20, 2025
+16%
Jul 27, 2025
+16%
Aug 3, 2025
+13%
Aug 10, 2025
+9%
Aug 17, 2025
+5%
Aug 24, 2025
+10%
Aug 31, 2025
+8%
Sep 7, 2025
+4%
Sep 14, 2025
+11%
Sep 21, 2025
+10%
Sep 28, 2025
+8%
Oct 5, 2025
+9%
Oct 12, 2025
+9%
Oct 19, 2025
+9%
Oct 26, 2025
+9%
Nov 2, 2025
+14%
Nov 9, 2025
+15%
Nov 16, 2025
+13%
Nov 23, 2025
+16%
Nov 30, 2025
+1%
Dec 7, 2025
+17%
Dec 14, 2025
+17%
Dec 21, 2025
+14%
Dec 28, 2025
-48%
Jan 4, 2026
-54%
Jan 11, 2026
+10%
Jan 18, 2026
+22%
Jan 25, 2026
+22%
Feb 1, 2026
+27%
Feb 8, 2026
+32%
Feb 15, 2026
+36%
Feb 22, 2026
+33%
Mar 1, 2026
+50%
Mar 8, 2026
+49%
Mar 15, 2026
+54%
Mar 22, 2026
+55%
Mar 29, 2026
+58%
Apr 5, 2026
+41%
Apr 12, 2026
+46%
Apr 19, 2026
+60%
Apr 26, 2026
+66%
May 3, 2026
+67%
May 10, 2026
+80%
May 17, 2026
+91%
May 24, 2026
+95%
May 31, 2026
+85%
Jun 7, 2026
+106%
Jun 14, 2026
+113%
Jun 21, 2026
+111%
More infoN = 47,900 paid workspaces (June 2026)
Output - Coding agents
Coding agents account for most of the acceleration
Teams that connected a coding agent roughly tripled their weekly pull requests over two years, from 21 to 65, while teams without one went from 8 to 10. These teams were already higher-output before coding agents existed, so the levels aren’t directly comparable, but each cohort against its own baseline tells a clean story, and nearly all the growth sits on the agent side.
Pull requests per team per week - Fixed cohort (paid workspaces)
706050403020100
Jul 2024Jan 2025Jul 2025Jan 2026Jul 2026
Average pull requests opened per workspace per week, coding-agent teams versus traditional teams, June 2024 to June 2026
Week of
Coding-agent teams
Traditional teams
Jun 2, 2024
21
8
Jun 9, 2024
24
8
Jun 16, 2024
24
9
Jun 23, 2024
22
8
Jun 30, 2024
24
9
Jul 7, 2024
21
8
Jul 14, 2024
24
8
Jul 21, 2024
24
8
Jul 28, 2024
24
9
Aug 4, 2024
24
8
Aug 11, 2024
24
8
Aug 18, 2024
23
8
Aug 25, 2024
25
8
Sep 1, 2024
24
8
Sep 8, 2024
24
8
Sep 15, 2024
25
9
Sep 22, 2024
25
8
Sep 29, 2024
26
9
Oct 6, 2024
25
9
Oct 13, 2024
26
8
Oct 20, 2024
25
8
Oct 27, 2024
26
9
Nov 3, 2024
25
8
Nov 10, 2024
27
9
Nov 17, 2024
26
8
Nov 24, 2024
28
9
Dec 1, 2024
23
8
Dec 8, 2024
27
9
Dec 15, 2024
28
9
Dec 22, 2024
26
8
Dec 29, 2024
10
3
Jan 5, 2025
11
4
Jan 12, 2025
25
8
Jan 19, 2025
27
9
Jan 26, 2025
27
8
Feb 2, 2025
28
8
Feb 9, 2025
29
9
Feb 16, 2025
30
9
Feb 23, 2025
29
9
Mar 2, 2025
30
9
Mar 9, 2025
30
9
Mar 16, 2025
30
9
Mar 23, 2025
31
9
Mar 30, 2025
31
9
Apr 6, 2025
30
8
Apr 13, 2025
32
9
Apr 20, 2025
28
8
Apr 27, 2025
28
8
May 4, 2025
28
8
May 11, 2025
29
8
May 18, 2025
32
9
May 25, 2025
31
9
Jun 1, 2025
28
8
Jun 8, 2025
31
8
Jun 15, 2025
31
8
Jun 22, 2025
30
8
Jun 29, 2025
32
8
Jul 6, 2025
29
8
Jul 13, 2025
32
8
Jul 20, 2025
31
8
Jul 27, 2025
32
8
Aug 3, 2025
32
8
Aug 10, 2025
32
8
Aug 17, 2025
31
7
Aug 24, 2025
33
8
Aug 31, 2025
32
8
Sep 7, 2025
31
8
Sep 14, 2025
34
8
Sep 21, 2025
34
8
Sep 28, 2025
34
8
Oct 5, 2025
35
8
Oct 12, 2025
34
8
Oct 19, 2025
34
8
Oct 26, 2025
34
8
Nov 2, 2025
36
8
Nov 9, 2025
36
8
Nov 16, 2025
35
8
Nov 23, 2025
37
8
Nov 30, 2025
31
7
Dec 7, 2025
37
8
Dec 14, 2025
38
8
Dec 21, 2025
37
8
Dec 28, 2025
16
3
Jan 4, 2026
13
3
Jan 11, 2026
35
7
Jan 18, 2026
40
8
Jan 25, 2026
39
8
Feb 1, 2026
42
8
Feb 8, 2026
44
8
Feb 15, 2026
46
9
Feb 22, 2026
44
8
Mar 1, 2026
49
9
Mar 8, 2026
50
9
Mar 15, 2026
51
9
Mar 22, 2026
50
9
Mar 29, 2026
52
9
Apr 5, 2026
48
9
Apr 12, 2026
49
9
Apr 19, 2026
54
9
Apr 26, 2026
55
9
May 3, 2026
55
8
May 10, 2026
57
9
May 17, 2026
60
10
May 24, 2026
62
10
May 31, 2026
57
9
Jun 7, 2026
65
10
Jun 14, 2026
63
9
Jun 21, 2026
65
10
Coding-agent teamsTraditional teamsMore infoN = 6,887 paid teams (4,280 with coding agents, 2,607 without)
A CLOSING NOTE
The clearest indication of AI’s influence on product development is the dramatic output gains experienced by teams using coding agents over the last two years. We have no way of knowing whether this increased output led to positive business outcomes, but it shows a very clear correlation between AI adoption and acceleration.
Perhaps more intriguing is the makeup of that adoption, and how it appears to be blurring roles. Senior leaders are doing more of the hands-on IC work, adopting AI aggressively to help them do it, and non-engineers are committing code. The suggestion that everyone in an organization is becoming a “builder” seems to be directionally true.
Those gains haven’t shown up as time saved, though. Time spent on existing tasks in Linear held while AI usage appeared as a new layer of work, meaning the overall time spent on product development is going up rather than down. As far as we can observe, teams are working more, not less, suggesting AI has a Jevons paradox quality beyond token consumption.
Many will rightfully argue that looking at pull requests indicates motion rather than value, which is certainly true, but it’s still a step forward from measuring tokens. A mechanical refactor might burn lots of tokens while a meaningful bug fix or code review doesn’t, so token spend and value don’t line up at all, and using one as a proxy for the other will be remembered as a relic of AI’s early days.
In future reports we intend to go deeper on the full lifecycle of work, from token spend all the way to outcomes, something we can newly observe now that code and code review run through Linear as well.
TIM QI - Head of data
Appendix
Methodology
This report uses aggregated product data from Linear. The data includes AI conversations, agent sessions, issue activity, comments, and pull requests. It covers only paid workspaces and the users in them. We report all metrics in aggregate to show broad patterns in how teams use AI to build software, not individual behavior. We measure each metric in a fixed time window. A window is one calendar month or the last 30 days. The year‑over‑year charts use June 2025 and June 2026. Adoption metrics use a trailing 30‑day window, and time‑series charts aggregate to weekly points. Both steps reduce short‑term noise. Some charts keep only the users who are active in both windows.
Definitions
AI-active. A user with at least one AI interaction, an in-app or Slack conversation or an agent session, in a 28-day window.
Agent team. A workspace with a coding agent connected.
Pull request. A code change opened against a repository connected to Linear. We count pull requests opened, not merged.
Paid workspace. A workspace on a paid plan, active during the relevant period.
Agent issue. This includes delegating an issue to an agent or starting a session.
Company size. Full-time employees at the company, from third-party enrichment.