JetBrains AI

The Brief

There's a particular kind of brief that arrives wrapped in panic. You can feel it in the room before anyone speaks: the slightly too-fast talking, the emphasis on competitor activity, the way questions about strategy get deflected into questions about timelines. JetBrains' brief was one of these.

The company had spent two decades building sophisticated development tools for professionals who cared about craft. IntelliJ IDEA, PyCharm, WebStorm; if you've never heard of them, then you've never worked in development, where they are ubiquitous and highly respected. Jetbrains builds premium tools for serious developers, and in doing so had built a loyal, almost tribal following among people who understood the difference between adequate and excellent.

Then the AI hypetrain arrived, and JetBrains found themselves dangerously late to the station.

Cursor and Windsurf, tools that barely existed a few years ago; were eating market share with AI-first approaches to coding. The industry narrative had shifted beneath JetBrains' feet. Suddenly 'anyone can code' wasn't aspiration but apparent reality. And JetBrains' marketing team, none of whom were developers themselves, were trying to figure out how to launch an AI offering into a market whose terms had already been set by competitors.

The Challenge

My first response to the brief was simple: don't.

I thought then, and still think now, that building AI coding tools was the wrong move for JetBrains entirely.

The company was sacrificing three things it couldn't afford to lose:

First, money: JetBrains was pouring resources into developing Jetti, Junie and Mellum (their own LLM) and building proprietary AI models from scratch; a fortune spent chasing a hype cycle when they could have simply built compatibility with existing AI tools. Let the infrastructure players burn cash on foundation models. JetBrains' value was never in the AI itself but in the developer experience wrapped around it.

Second, credibility: The developer community was deeply divided on AI, and a significant proportion felt genuinely conflicted about tools they saw as threatening their livelihoods. The coding knowledge embedded in these models came directly from datasets scraped from Stack Overflow and GitHub; developers were essentially being sold their own expertise back to them, often without attribution. Even developers who used AI tools regularly described feeling 'guilty' or 'embarrassed' about it. JetBrains, a company built on developer trust, was wading into ethically contested waters.

Third, a massive differentiation opportunity: In a market where every player was racing to out-AI each other and burning through cash on comms, the marketplace was noisy, crowded and impossible to cut-through without VC backing that Jetbrains simply do not have. Instead, JetBrains had the heritage and credibility to stake out entirely different ground. They could have been the brand that stood for developers; that publicly questioned the 'AI will replace programmers' narrative; that positioned itself as the tools for people who believe human expertise still matters. Instead of chasing Cursor, they could have been the antidote to Cursor. The PR potential alone was enormous: the twenty-year-old developer tools company willing to say what everyone was thinking but no one in the industry would say out loud.

I was overruled. JetBrains was committed to launching an AI product; the market pressure was too intense, the board too anxious. And with the benefit of hindsight, as the AI bubble shows early signs of deflation and JetBrains' AI offerings struggle to gain meaningful traction against entrenched competitors, I remain convinced the contrarian path would have served them better.

Still, the brief is the brief. But the challenge shifted somewhat:

How do you launch an AI product for a company that shouldn't be launching an AI product? How do you thread a needle between market expectations and brand integrity?

The answer required going deeper than messaging. I needed to understand what was actually happening in software development as a result of AI democratisation; and whether there was a genuine problem that JetBrains' particular heritage positioned them to solve.

The Three Horsemen of the Slopocalypse

The democratisation of coding through AI had led to an unprecedented flood of software. The numbers were staggering: 41% growth in AI repositories on GitHub,126% more project completions with AI assistance, 252,000 new websites daily. This was unquestionably good in many respects; more people able to build things, lower barriers to creation, faster iteration.

But like all technological shifts, it came with downstream consequences that were only beginning to emerge. I identified three.

Vibeware. Software created based on fleeting inspiration or trending ideas rather than genuine user need. Built rapidly using AI tools with minimal consideration for purpose, utility, or long-term value. The app stores were drowning in it: 62,000+ new apps published monthly, 1.14 million apps on Apple's App Store that had never received a single user rating. The vibeware cycle was depressingly predictable: see trending concept, generate with AI, ship immediately, move to next trend. The result was a digital landscape full of functional but forgettable software solving problems nobody actually had.

Sprint Rot. The degradation of code quality that occurs when AI-enabled rapid development prioritises velocity over structure. GitClear's research showed an 8-fold increase in duplicated code blocks during 2024, a 40% decrease in refactored code, and for the first time ever, copy-pasted lines exceeding moved lines. This strongly implies that developers were accepting AI-generated code wholesale without taking the time to understand its architecture or implications. The tools enabling even experienced developers to build so quickly were creating codebases that would become maintenance nightmares over the long term. which brings us to…

Technical Debt. This is the thing that keeps serious devs and CTOs up at night; the implied cost of additional rework caused by choosing quick, expedient solutions over better approaches. In the AI era, this debt is accumulating faster than ever. One veteran developer I quoted in the presentation put it starkly: 'I don't think I have ever seen so much technical debt being created in such a short period of time during my 35-year career in technology.'

These three problems are the real cost of the 'anyone can code' revolution. Yet they point toward something JetBrains is uniquely positioned to address.

The Reframe

The strategic insight was this: everyone else was selling AI that enables anyone to build anything. JetBrains should be selling AI that helps developers build right.

This isn't branding sophistry so much as it is a genuine reflection of philosophical difference. ChatGPT and Claude were designed for generalists. Cursor and Windsurf were built around speed and accessibility. But JetBrains had spent twenty years building sophisticated tools for developers who cared about craft. Where the newcomers democratised coding for everyone, JetBrains had always been about premium tools for professionals who understood the difference between 'it works' and 'it's right.'

This heritage made JetBrains the natural champion of thoughtful development over thoughtless creation; a positioning that mass-market AI tools couldn't credibly claim.

The proposition I developed: Build Right.

Not 'build faster' (everyone claims that). Not 'build more' (that's the problem). Build right. Build with skill, with creativity, with foresight, with responsibility. Build software that endures rather than software that ships.

The competitive reframe was sharp: Competitors say 'anyone can code now.' JetBrains says 'developers always could; now they can do it better.'

What This Required

The strategy demanded that JetBrains stop chasing the democratisation narrative and instead position themselves against it. Not against AI itself; they were launching an AI product, after all; but against the quantity-over-quality culture that AI democratisation was creating.

This meant speaking to developers who believed there's a difference between 'can build anything' and 'should build this.' It meant appealing to technical decision-makers who were watching technical debt accumulate across their organisations. It meant owning the artistry: the taste, skill, creativity, and responsibility that separates good AI-generated code from great code.

The emotional benefit for individual developers: finally, AI tools that respect and build on your expertise rather than treating you as an obstacle to automation. The rational benefit for enterprises: reduced maintenance costs, consistent code quality, development that scales with intention rather than just velocity.

The message flexed naturally across audiences. For junior developers: 'Learn to be a developer, not just someone who uses AI.' For enterprise buyers: 'The AI development platform that builds your competitive advantage, not your long-term costs.' For hiring managers: 'Attract developers who care about craft, not just speed.'

The Bigger Picture

JetBrains ultimately went a different direction, but the work crystallised something I keep returning to in AI strategy.

The dominant narrative frames AI capability as inherently good: more power, more speed, more output. The assumption is that barriers to creation are simply obstacles to be removed, yet creation without intention produces noise; speed without craft produces debt; democratisation without standards produces vibeware.

The companies that will build lasting value in the AI era won't be those that enable the most output. They'll be those that enable the right output; that understand the difference between capability and quality, between what technology makes possible and what's actually worth building.

JetBrains had the heritage to own that position. Whether they ultimately claim it remains to be seen.

This strategy was developed for JetBrains in 2024-2025 as part of their AI product launch.