TL;DR
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
Linear co-founder Karri Saarinen argues that software reflects the experience, judgment and priorities of the people who make it. In an essay, he warns that AI-assisted development can make it easier to add features without shaping them into a coherent product, while saying faster tools could also give makers more room to refine their point of view.
Karri Saarinen, co-founder of software company Linear, argues that a product’s character comes from the experience and judgment of the people building it, not just its features or development process. In an essay titled Someone behind the software, he connects that view to the growing use of AI agents, warning that faster, more independent changes can make products feel like collections of capabilities rather than coherent answers to users’ problems.
Saarinen says two teams can speak with similar customers and use the same technology yet produce different products because they interpret what they learn differently. A company’s founders and culture also shape those choices. He describes Linear’s approach as valuing intuition over process or data, while clarifying that he does not mean decisions based on unsupported personal taste. In his account, judgment develops through experience, customer conversations, mistakes and a clearer understanding of what deserves attention.
He says Linear’s team spoke with people in different roles and companies before launching, using those conversations to test and improve its own perspective. Product decisions, he argues, include not only which features to build but what to leave out and what to keep refining after it works technically. A feature can function correctly on its own and still make the wider product worse, he writes, making coherence and judgment part of the work rather than an automatic result of adding functionality.
Saarinen says he has recently found products from some successful companies increasingly messy, with changes focused on what technology can do and what else can be added rather than on the shape of the product or its core problem. He suggests that additive work is often easier than rethinking existing capabilities. With AI agents enabling individuals to produce changes independently, he worries that contributions may make sense in isolation but weaken the shared direction of the whole product.
Why Product Judgment Matters
Saarinen’s argument addresses a practical question for software teams adopting AI tools: faster production does not by itself establish product quality. If separate contributors generate or implement changes without a shared understanding of the product, the result may contain more working features but offer a less consistent experience. That concern is relevant to users because the quality of software depends not only on whether individual functions operate, but on whether they fit together and address a real need.
His essay is a perspective from a company founder, not a reported study measuring how AI-assisted development affects product quality. It does not establish that agent use causes software to become less coherent. Rather, Saarinen identifies a risk in organizing work as a set of independent outputs and argues that teams still need people who can make product-level decisions, stay close to customers and revise those decisions when evidence changes.
He also presents a possible upside: if tools make implementation easier, makers could spend more time understanding problems and refining solutions. That outcome is not guaranteed by the technology. It depends on how teams use the time and capacity the tools provide, and whether they preserve a clear point of view while adding capabilities.
software development project management tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
How Saarinen Defines Intuition
Saarinen distinguishes informed intuition from making decisions by preference alone. In the essay, intuition is accumulated judgment: learning through building, talking to customers, making mistakes and noticing which details matter. He says a maker’s perspective can improve through outside conversations, rather than being limited to the founders’ initial frustrations or assumptions.
The essay places this view against a familiar model of software work as a repeatable pipeline: gather requirements, build features and measure results. Saarinen does not argue that customer input or data are useless. His point is that those inputs still require interpretation. Teams make choices about what problems to prioritize, which requests to decline and how separate functions should fit together.
He uses the idea of a product’s “soul” as a metaphor for that accumulated set of decisions. The term refers to a sense that someone has shaped the whole experience, not to a measurable product property. His concern about agent-supported work follows from that framing: if changes are made in separate streams, the common perspective that gives them direction may become less visible.
“Good products carry something of the people who make them.”
— Karri Saarinen, Linear co-founder
As an affiliate, we earn on qualifying purchases.
What the Essay Does Not Measure
The essay offers Saarinen’s perspective and examples of how he thinks about product design; it does not name specific products he considers messy or provide data showing that software quality has declined. He does not quantify how much agent use is taking place, compare agent-assisted and conventional teams, or show that AI tools are responsible for the product changes he describes.
It is also unclear how teams should measure whether a product has a coherent point of view, or how they should balance founder judgment with customer evidence and broader organizational input. Saarinen’s account does not detail particular changes at Linear or provide an outside evaluation of the company’s process. His essay therefore raises questions about development practices rather than establishing a general finding about the effects of AI agents.
As an affiliate, we earn on qualifying purchases.
How Teams Apply the Argument
The source material does not announce a product launch, policy change or research project, and it gives no scheduled next milestone. The immediate development is the publication of Saarinen’s argument. Any further test of the concerns he raises would require details about how teams use agents, how they review changes across a product and what outcomes they track for customers.
For readers following software development, the practical question is whether companies using AI-assisted tools can keep independent contributions aligned with a shared product direction. Saarinen’s essay points to customer contact, deliberate choices about what not to add and continued refinement after a feature works as parts of that effort. It remains to be seen whether those practices will change as agent-supported development spreads.
As an affiliate, we earn on qualifying purchases.
Key Questions
Who wrote “Someone behind the software”?
Karri Saarinen, a co-founder of Linear, wrote the essay. It presents his views on how makers’ experience and decisions shape software products.
What does Saarinen mean by a product’s “soul”?
He uses “soul” as a metaphor for a product’s coherent character: the sense that its features and details have been shaped by a shared perspective and care for the whole experience.
Is Saarinen saying software teams should ignore data?
No. He says intuition develops through experience, customer conversations and learning from mistakes. His argument is that information still needs interpretation and that teams should not hide from making judgments behind process or data alone.
Does the essay prove AI agents make products worse?
No. The essay does not provide measurements or comparative evidence. Saarinen describes a potential risk: independent changes may fit together poorly if teams lack a shared product perspective.
What does Saarinen hope easier software development will allow?
He hopes that easier production will give makers more room to understand problems and keep refining solutions until the parts belong together. He does not claim that faster tools will produce that result automatically.
Source: hn
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
