Linear optimizes CI pipeline as AI agents accelerate code shipping
The development tool company reworked its continuous integration infrastructure to handle code validation at the pace AI coding assistants now generate changes.
What to know
- AI coding assistants have become so fast at generating code that continuous integration—the automated testing and validation stage—is now the constraint slowing down the development cycle, forcing infrastructure investment.
- Linear cut pull request validation time to just over 5 minutes through three approaches: moving to faster third-party CI runners (34% speed gain), adopting a native TypeScript compiler (73% faster type-checking), and refactoring lint rules to avoid expensive type-graph building.
- The optimization reflects a broader tension in AI-accelerated development: tools ship code faster, but the validation, testing, and human product decisions haven't accelerated equally, raising questions about whether speed gains translate to better products.
The dispute Whether AI's measured productivity gains at the infrastructure level (faster CI, more code shipped) translate to meaningful improvements in end products, revenue, or user experience. · positions read across 46 posts and comments
AI is driving tactical improvements in infrastructure and process automation, though not yet in final product outcomes.
-
“A good chunk of what my company has been doing with AI falls into either burning down our known tech-debt and "easy wins" that no one ever had the bandwidth to approach... And improving / automating our processes.”
TrueDuality · Hacker News ↗
Speed gains don't matter if validation and product decisions remain slow; CI optimization alone doesn't solve the real bottleneck.
-
“In my case it's not the CI that's the bottleneck. It's the human testing side. Does it work, sure. But does it actually do the thing we want?”
aliclark · Hacker News ↗
Despite AI acceleration, visible product improvements are lacking and fundamental OS/platform issues remain unresolved.
-
“While every post and thread feels like a 90's wall street office, the new android and iphone ship with fewer features than usual. No indie guys come up with a linux-sized alternative OS. Switch 2 remains unhacked. Windows takes 3 seconds…”
torben-friis · Hacker News ↗
Linear Software development tool companyTuomas Linear CTOjulian_digital Linear engineer
How it unfolded 2 developments, newest first · click a bar or a number to jump articlespostscomments
-
2
Companies report AI helping with tech debt and process automation, not just speed
A commenter noted that their company's AI use is focused on burning down known technical debt and automating QA processes, allowing QA teams to shift from blockers to proactive reviews and deeper issue discovery.
“A good chunk of what my company has been doing with AI falls into either burning down our known tech-debt and "easy wins" that no one ever had the bandwidth to approach... And improving / automating our processes.”
— TrueDuality -
because the tooling is struggling to catch up and adapt. there's still no vetted all-in-one solution that does code review, charts quality metrics, ingests incidents, and writes tests. there could be but each solution would necessarily need to be so bespoke to a codebase that you might as well hire some AI-pilled QAs and come up with it yourself…
2 more of the top 3 · 40 posts in this stretch
-
A good chunk of what my company has been doing with AI falls into either burning down our known tech-debt and "easy wins" that no one ever had the bandwidth to approach... And improving / automating our processes. The former is having a direct and meaningful impact on the quality and availability of our services.Our QA, formerly a fairly frequent…
-
It's the difference between the value of "code" vs "software". Lots of people use those phrases interchangeably but its becoming painfully apparent that doesn't work anymore. Software requires more effort to actually make something IRL. Code can just be generated and sit in a repo, doing nothing, interacting with nothing.Github has reported a 14x…
-
-
1
Developer questions whether speed gains translate to product improvements
A commenter raised skepticism about whether AI productivity gains are yielding visible improvements in shipped products, noting that new OS releases ship with fewer features than usual and foundational problems persist.
“While every post and thread feels like a 90's wall street office, the new android and iphone ship with fewer features than usual. No indie guys come up with a linux-sized alternative OS.”
— torben-friis -
Here's my constant question:Everyone's going so fast that they keep hitting walls. Review, CI, product asking for things, whatever.Why have we not seen an improvements in products?While every post and thread feels like a 90's wall street office, the new android and iphone ship with fewer features than usual. No indie guys come up with a…
2 more of the top 3 · 6 posts in this stretch
-
> Moving our workloads off GitHub Actions to third-party runners with faster CPUs, higher-performance storage, and better cache infrastructure gave us faster machines to run the same pipeline onYeah, was not surprised to read this. Actions is convenient if you already use GitHub, but it can also be pretty slow. Given reliability is also a major…
-
My take: it seems like systems should become smaller, more isolated, and contract-oriented.I have been a long time proponent of monoliths, but it seems like agents would be happier with smaller, more isolated services. The more isolated, the better. Contracts between the service components only. Then it can iterate internally as long as it…
-
-
background
Linear optimizes change-detection jobs by limiting fetch depth — The company capped fetch depth and removed unnecessary checkout operations from jobs that didn't need them, reducing the slowest change-detection gate from 94 seconds to 20 seconds.
-
background
Linear rewrites lint rules to eliminate TypeScript type dependency — Custom lint rules were refactored to use static analysis on the abstract syntax tree instead of building the full type graph, letting ESLint drop TypeScript entirely. This reduced API lint time by 68% and full-repository lint time by 55%.
-
background
Linear switches to native TypeScript compiler, cuts type-checking by 73% — Switching to tsgo, a native TypeScript compiler, cut the median weekly tsc check by 73%, moving the bottleneck off typechecking entirely.
-
background
Linear moves workloads off GitHub Actions to third-party runners — Moving CI workloads from GitHub Actions to third-party runners with faster CPUs and better cache infrastructure delivered a 34% average speed improvement, with some compiler checks like tsc dropping 52%.
-
background
Linear publishes CI optimization retrospective with measurable gains — Linear published a detailed account of its CI rework. Despite test suites nearly quadrupling since the start of the year, the company reduced pull request wait time from over 6 minutes to just over 5 minutes and cut runner time per test roughly in half.
-
background
Linear CTO assigns CI optimization task amid cost and speed concerns — Tuomas, Linear's CTO, assigned an engineer the issue "CI costs are high" and requested faster CI. The trigger reflected a broader shift: AI agents had made code shipping exponentially faster, but validation pipelines had not kept pace.
Also covered reported alongside — the timeline has no entry for these yet
-
first by HN Best, 2d ago · also HN Frontpage
What people are saying 18 voices from 1 site · best of 46 · verbatim
- If AI is making development so much faster, where are the tangible improvements in shipped products and user experience?
- Why are companies able to accelerate code shipping but not seeing commensurate revenue growth?
- Sep 22
-
go and look at how the good people and claude over at rsync are fixing all their bad security bugs currently: https://github.com/RsyncProject/rsync/commit/42f7c32ce51d8e0... Also: any Linux-distro-bugtracker...So: while running in circles might be a bit of an exaggeration, it seems (which maps to my experience) these things are much less AGI than…
-
There won't be any improvement in products. Most of software development is kayfabe performed for clueless executives. Meaningless and worse than useless software being developed as something new, because new gets promotion, bonus. Everyone is always developing something new and sometimes 3 - 6 new solutions by different orgs at the same time for…
-
I’d love to see data. For my part the products I use seem to be moving faster, with more small fixes, accessibility better designed in, localization better (even if imperfect) and more rapid releases.I’m not sure Google, Apple, and Microsoft are the best barometers of AI impact on software engineering. They write software, but at such scale and…
-
>Why have we not seen an improvements in products?This was never the goal, because it would influence the standards and expectations of the end user. Any gamer remembers when Baldur's Gate 3 came out everyone said somewhere between "this is the new standard in AAA games" or "this is an amazing value for money". The response from the rest of the…
-
Moose has entered the room.I built a debian based OS focused on self-hosting: github com/onmoose/os It's not really Linux size but I couldn't have built this if it wasn't for coding agents. Still a ton of work needed on UX but the main functionality works so well.Linux also has decades of work on it, we can't really expect new products to equal…
-
It really depends on the industry.The blind community is benefiting enormously from coding agents. Game accessibility mods for everything under the sun (the big names in the last month or two are Civ V and Witcher, although there's plenty more), accessible 3rd party clients for annoying sites, people's favorite speech synthesizers ported to…
-
Personally I’m leveling up on using AI to develop a project. My speed at knocking out PRs has increased a lot, but much of that is massaging the process of using AI, including managing CI bottlenecks and costs. I’ve never had to deal with 15+ PRs being developed in parallel before. It’s interesting work, but there’s a lot of doing stuff to get to…
-
>Because the third-party runners sit outside GitHub’s network, they rely on a direct IP link to reach GitHub. The provider traced the hangs to intermittent degradation on that link.This is interesting. If I read that correctly this means the CI host had a direct peering connection with Github or Azure network.Why were they not able solve the…
- Sep 21
-
Probably for the same reason that SV companies hiring thousands of developers struggled to improve their product much past the original product, that was built by a handful of people.Scale in headcount was a tactic to get investment, then you had to find stuff for everyone to work on. Suddenly people have the time to engineer so hard that we get…
-
Agreed, but give it time. The tipping point of good models only just arrived with Fable 5, GLM 5.3, Grok 4.6, Muse Spark 1.3, and the like. Those are the only models that do a better job than me, and I'm happy to hang up my IDE—and that was barely a month ago.I've been fixing everything that has been sitting there—not blocking us, but slowing us…
-
incentives are not aligned to improving products.the important thing to microsoft is still finding new ways to make money, not to make your clicks faster. they might even be using their new velocity to make clicks slower with an ad in between.there's alao much less incentive to build things to share. people can just make their own widget app and…
-
PS5 emulation has gone from barely working to running Dark Souls at 10+ FPS with virtually no graphical glitches… in 6 weeks. If you’re not familiar with normal emulator development time, this is… quite extraordinary. There have been insane progress on decompilations and many other things in the emulator space.Everyone’s trying to figure out how…
-
I think to solve this problem properly, we have to stop treating "build" and "test" as separate buckets of work, to be designed and scaled separately. It's all CI. As soon as you try seriously scaling out tests, you will run into build bottlenecks. To truly scale CI you need a scheduler that understands your build, test environment, and all the…
-
I suspect a substantial part of this is an avalanche of useless testing.If you even review PRs still: when was the last time you didn't just skip over tests? And if you ever looked at tests in an LLM-heavy PR, how many of those tests tested something useful, and not just built-ins and trivial behaviours?There's at least some awareness in the…
-
> Why have we not seen an improvements in products? [..] Is everyone just running full speed in circles or something?The simplest explanation is that they don't give a flying flamingo about what you or I consider "improvements to products".This report is an example.There are several changes that modify CI behaviour, where the article gives no…
-
With the widespread increase in speed/commit cadence/content added to repos, I wonder if there's an angle here for companies with a lot of CI/CD needs to simply start self-hosting their CI/CD machines on premises.In my experience, self hosting your CI/CD runners had the biggest impact in cost savings throughout, while also allowing for more…
-
My biggest takeaway (as a one man band 20 leagues below this spending time optimizing Gitlab free plan and self hosted runners): they did not worry about any of this until 100M ARR and more 1B valuation[0].[0]
-
Last year people were asking "if AI is so great then where are the new apps?". Then data for 2026 came out and now the IOS app store has a 84% percent year-over-year increase in new app submissions.For the question where are the alternative OSes? Here is one that I've seen. There's probably more -…