# If SaaS Is Dead, Linear Didn't Get the Memo

Podcast: AI & I
Published: May 5, 2026
Reading time: 22 min
Canonical: https://podbrew.app/briefs/ai-and-i-if-saas-is-dead-linear-didn-t-get-the-memo

Dan Shipper talks with Karri Saarinen, cofounder and CEO of Linear, a product management tool. Their discussion focuses on Linear's successful adaptation to the AI era, transforming into an agent-native business. They examine how a company founded before ChatGPT reinvented its approach to software development.

The conversation details building a platform where humans and AI agents collaborate on software development. It also scrutinizes the widely discussed idea that "SaaS is dead," arguing that this narrative oversimplifies the true impact of AI on the industry. Listeners gain insight into the nuanced future of software as a service.

This episode is important because it offers a real-world example of an established company navigating the complexities of AI integration. It provides valuable perspectives on how businesses can evolve, leverage new technology, and rethink product strategy in a rapidly changing technological landscape. The discussion challenges assumptions and offers practical insights into the future of work and software.

## Key takeaways

- Linear embraced a patient, craft-oriented approach to product development, prioritizing quality and thoughtful design by initially limiting access and avoiding excessive external pressures.

- Prioritize understanding user workflows over rushing to implement popular new technologies, as evidenced by Linear's delayed adoption of early AI chatbots.

- An open platform strategy, allowing third-party agents and custom-built solutions to integrate, can be more effective than attempting to build an all-encompassing AI solution.

- Linear acts as a system for guiding AI agents and providing context, helping organizations decide what agents should work on to ensure productive and useful outcomes.

- Resist market pressure to integrate new technologies prematurely; prioritize finding real value and solving actual problems.

- The "SaaS is Dead" narrative is an oversimplification; the biggest disruption will likely hit large, inflexible public companies, not necessarily eliminate entire software categories.

- Focusing on the quality and value generated by AI-assisted work is more important than tracking vanity metrics like the volume of AI-written code.

- Relying on simple metrics like AI token usage or PR count is misleading, as activity does not always equal positive impact in product development.

- Success in AI development should be measured by classic business metrics like profits, revenue, and user satisfaction.

- Linear enforces a "zero bugs" policy, requiring all reported bugs to be fixed within a one-week SLA.

- AI coding agents are integrated into the workflow to perform initial bug fixes, with engineers reviewing and refining the solutions in Glanair.

- Linear's AI synthesizes numerous customer feature requests to pinpoint core problems, significantly accelerating the initial understanding phase of product development.

- AI-powered integrations, such as a Slack agent creating issues directly from conversations, streamline workflows by transforming discussions into actionable tasks immediately, shortening overall development cycles.

- Teams should dedicate time to thoroughly identify the right problem and its optimal approach before committing to a solution.

- Once a deliberate commitment is made to a project, the subsequent execution phase should be fast and efficient.

- Building an in-house coding agent allows for deeper integration and leverages existing product context, enabling a more seamless end-to-end workflow.

- Integrating proprietary AI agents enables tighter feature control but introduces significant operational costs for the company due to token usage.

- High-usage AI features, particularly coding agents, necessitate a usage-based billing model to manage potentially very expensive operational expenses.

- Linear uses custom AI agents, trained on internal company data, to act as product teammates for synthesizing complex user problems.

- Product building remains a human craft driven by intuition and problem understanding, where data serves as a signal rather than the primary decision-maker.

## 00:00 - 04:01 Linear's Patient Philosophy and AI Transition

Linear, co-founded by Karri Saarinen, developed its product with a philosophy centered on patience and craft, initially keeping access limited and avoiding high funding expectations. This approach fostered a reputation for quality, allowing the company to build a strong foundation over the long term, admired for its thoughtful execution and high taste.

Despite initially not rushing into AI when GPT-3 first emerged, Linear successfully transitioned its product to be "agent-native" in the current AI era. This adaptation has positioned Linear as a standout example among pre-AI companies successfully integrating new technologies, even serving as a key integration point for products like OpenAI's Symphony.

For Linear, the core mission of helping companies move work forward and build software has remained consistent, even as AI reshapes the landscape. The company views AI as a tool that enhances its mission by automating more burdens, thereby allowing product teams and individuals to focus their energy on creative craft, taste, and deeper thinking rather than administrative tasks.

> In some ways, the mission for us didn't change. The AI is making it better because now we can automate more and take more of that burden and let people use their craft or taste or thinking.

## 04:01 - 08:02 Linear's Design-First Approach to AI Workflows and Open Agent Platform

Coming from a design background, Linear's co-founder Karri Saarinen emphasizes the importance of deeply understanding a problem before jumping to solutions, a contrast to many tech companies that rush to implement new features. This philosophy guided their cautious approach to early AI chatbots, where they realized internally that simply adding a chatbot wasn't useful without a clear understanding of the user workflow it would serve.

Instead of a generic chatbot, Linear spent years researching how people would genuinely want to use AI. This led them to develop an open agent platform with robust documentation, enabling various coding agents, including OpenAI's Codex, to integrate seamlessly with Linear. This strategy acknowledges that there won't be a single dominant AI agent, but rather many, with companies building their own specialized agents that can interoperate.

Linear's role evolved into a system for guiding these diverse agents and providing the necessary context. They now offer a chat interface within Linear, designed not for general chat, but for specific, valuable tasks like synthesizing customer requests. The goal is to bring clarity and context to organizations, helping them productively harness AI by focusing agents on genuinely important tasks, rather than just any task they can perform.

> How do you actually use-- like, what is the workflow where you would actually need this or use this?

## 08:02 - 12:02 The "SaaS is Dead" Narrative is Simplistic, Requiring a Day-One Mindset

Companies often face pressure, both internal and external, to implement new technologies like AI even before their true value is understood. Rather than rushing to release AI features, it's crucial to identify actual problems that the technology can solve, focusing on tangible benefits for users and businesses. This patient approach avoids adopting fads.

The rapidly evolving AI landscape generates significant market noise, with trends and best practices changing frequently. Many emerging AI concepts are untested, particularly in large organizational settings where their efficacy truly matters. It's essential to analyze these signals critically and avoid making hasty predictions or adopting unproven solutions.

The "SaaS is Dead" narrative, suggesting a complete overhaul of existing tools like CRMs, is too simplistic. While the landscape is indeed shifting and introduces uncertainty for future cash flows, the most significant impact will likely be felt by large, less flexible public companies. Their established moats may diminish, rather than entire categories of software being wiped out.

Even established companies must adopt a "day one" mindset, continually re-evaluating problems and solutions from a fresh perspective. This means not relying on past product decisions or existing offerings, but rather considering how new technologies, like agents in product development, fundamentally alter problems and create opportunities for new solutions.

> We need to live in this day one world again, where like we can't rely on our previous decisions anymore, like we have to like look at these problems like in a fresh way.

## 12:02 - 14:02 Linear's Transition to Widespread AI Coding Tool Adoption

Linear faced the common challenge of integrating AI coding tools, overcoming initial skepticism among its product team, which consists of about 60 people including engineers, designers, and product managers. There was initial reluctance and questions about whether AI was just glorified autocomplete or a threat to programming jobs.

The company actively encouraged its team to adopt these new tools, recognizing that existing habits often make people less inclined to try new workflows. This involved fostering an environment where trying new approaches was welcomed and supported.

The effort paid off, as now nearly all of Linear's engineering team, along with some designers and product managers, are regularly using AI coding tools. This widespread internal adoption demonstrates a successful shift in workflow and mindset.

Linear deliberately avoids tracking vanity metrics like the percentage of code written by AI. Instead, the focus is on measuring the actual value and quality generated by the output. The company emphasizes that simply generating more code with AI does not automatically equate to improving the product or providing meaningful work.

> We had to encourage people to use these tools more. I think there's always that, there can be habits where you've always done stuff this way, so you are kind of less and less interested in trying new tools.

## 14:02 - 16:02 Prioritize Classic Business Metrics Over Simplistic AI Usage

The current market structure incentivizes large companies, often token sellers, to promote increased token consumption, leading to a simplistic belief that higher token usage equates to better outcomes. However, this approach overlooks that activity in product building isn't always positive; changes or additions can sometimes have negative impacts, making raw token spend an unreliable indicator of success.

Instead of focusing on vanity metrics like the number of AI tokens used or pull requests submitted, the primary focus should remain on classic business metrics. These include profits, revenue, and user satisfaction, which represent the ultimate goals of any product development effort. While these are often lagging indicators, they provide a more accurate measure of true value generation.

Although specific AI-related metrics, such as token usage per person or by different teams, can serve as internal signals of activity, they should not be treated as the sole or primary measure of success. These metrics can indicate whether "we are doing something," but they do not inherently answer whether that "something" is effective or contributing positively to the product and business objectives.

> I think it's still the classic metrics of like, profits or revenue or user love or some of these things are like what you should be aiming for.

## 16:02 - 18:02 Linear's Zero Bugs Policy and AI-Powered Bug Fixing

Linear maintains a strict "zero bugs" policy, where any identified bug must be fixed within a one-week Service Level Agreement (SLA). This commitment to quality is seen as a measurable metric of product improvement, moving beyond simply increasing output to focusing on the actual user experience.

The company leverages AI coding agents to streamline this process. These agents perform the initial pass on bug fixes, significantly accelerating the early stages of resolution. Once the AI generates a fix, it alerts an engineer for review.

Engineers can then review the code and make any necessary changes directly within Glanair, ensuring human oversight and refinement of the AI-generated solutions. This integrated workflow enhances efficiency while maintaining the high quality standards Linear expects.

This approach stems from a deliberate choice to prioritize product quality. Linear views bugs as mistakes that should be rectified promptly, establishing bug fixing as a top priority for everyone involved, rather than just seeking more output.

> we think bugs are like bad things or mistakes, and we should fix them as quickly as we can, and that's a priority to everyone.

## 18:02 - 22:04 Linear's AI Synthesizes Customer Requests and Shortens Development Loops

Linear leverages AI to synthesize customer feature requests, analyzing feedback from hundreds of users to quickly identify the core problem. This capability helps product managers understand new requests efficiently, informing decisions on whether a feature should be prioritized for development immediately or postponed, without requiring extensive manual investigation.

While manual exploration in design tools like Figma is still valued for initial concept generation, Linear's robust build system supports rapid prototyping for the product and design teams. This system allows for the creation of pull requests that generate live preview links, significantly streamlining the testing and iteration phases of feature development.

On the engineering side, Linear integrates with communication platforms like Slack, where its agent can automatically convert team discussions into actionable issues. This eliminates the need for separate meetings or manual task creation, allowing teams to move from discussion to execution almost instantaneously.

The primary benefit across these workflows is the substantial shortening of various development loops. By automating the synthesis of information and the creation of actionable tasks, Linear enables teams to respond to problems and implement solutions much faster, reducing delays inherent in traditional product development processes.

> I think it's kind of like, I, I would say like, kind of like the pattern in all of those things is like it's shortening the, some kind of loop there.

## 22:03 - 24:03 Prioritize deliberate decision-making before accelerating execution

Rushing through the decision-making process can lead to building products or features without a clear understanding of their purpose or value. Some teams develop ideas quickly, resulting in prototypes that might seem useful but lack a well-defined framework for assessing their true impact or comparing them against other priorities.

This creates a risk of committing time and resources to projects that haven't been thoroughly vetted, making it difficult to justify their existence or allocate further development time effectively. The initial rush can lead to a state where an idea is built, but no one truly understands why it exists or whether it should have been pursued in the first place.

Linear's approach emphasizes a deliberate commitment to an idea, fix, or project before accelerating execution. While the company doesn't rely on rigid processes for decision-making, the focus is on a clear commitment. Once a team commits to a project, the goal is to make the execution loop as fast as possible to efficiently work on the problem.

The core distinction lies between problem-finding and problem-solving. Teams should take the necessary time to find the right problem and determine the most suitable approach. Only after this thoughtful commitment to the problem and its solution has been made should the execution phase be accelerated.

> I don't want the problem finding to be fast, like you should take the time to find the right problem and like the right approach for the problem, and then once you decide that, then you can go faster on it.

## 26:03 - 30:04 Focus on Conceptual Work Before Shipping Solutions

Conceptual work in design and product development should be treated like building a "concept car." The initial output is not a product ready for production, but rather a collection of ideas intended to explore possibilities and deepen understanding of a problem. This approach allows teams to evaluate the merit of a new idea or a drastically different approach to a surface, without immediately confronting the fears and complexities of implementation changes.

The aim of this internal exploration is not always to ship a product immediately. Sometimes, the most valuable output is a better internal understanding of a problem. This enhanced clarity then allows the team to tackle the problem more effectively and eventually create a shippable product. Building can be part of this thinking process, but the purpose is learning and understanding, not immediate customer deployment.

After internal conceptual work, products move into customer-facing betas. It's crucial to define clear goals for this stage. The objective should be to understand user workflows and how people interact with the product, focusing on how it can be improved. The goal is not to ship the product as fast as possible, but to gather insights from real-world usage to guide further development.

Even with modern tooling and AI, there's a limit to how much internal thinking can happen before external validation is needed. The process involves stages, and each stage requires an honest assessment of its specific goals, ensuring that teams are focused on learning and refining rather than prematurely pushing a solution to market.

## 30:04 - 32:05 Linear's AI Agents Transform Product Strategy Beyond Simple Ticketing

Linear is shifting its product strategy to move beyond the traditional understanding of issue tracking as a mere ticketing system. Instead, the company views its platform as a fundamental "backbone" for gathering critical organizational signals, identifying problems, and tracking key decisions.

New AI-powered tools are central to this evolution. A "Linear agent" provides contextual support for product managers and designers, helping them understand issues. Furthermore, a "coding agent" is being developed to assist with writing code, allowing users to see diffs online in a "cloud conductor" environment.

This strategic pivot aims to significantly enhance team workflows. By automating routine tasks typically associated with a "kitchen ticketing system," Linear empowers its users to focus on higher-value activities like shaping work and leveraging the comprehensive context the platform collects.

> for us, like, Liner is more like the backbone we rely on, Collecting signals and collecting problems or collecting decisions like we should do this thing.

## 32:04 - 34:05 Linear Builds In-House Coding Agents for Seamless End-to-End Workflows

Linear decided to develop its own coding agent internally, recognizing a significant limitation with external AI tools like Claude or ChatGPT. When using these third-party agents, users are constantly required to explicitly feed them context for every task, leading to an inefficient and fragmented workflow.

By contrast, Linear's in-house agent leverages the rich context that already resides within the Linear platform. This inherent knowledge base allows for a much smoother, end-to-end experience where tasks can be initiated, issues created, code written, and diffs reviewed and merged without constant manual context input.

This integrated approach enables a more natural and automated workflow. Instead of spamming context windows, the Linear agent can intelligently utilize existing project and organizational data to automate common tasks such as bug fixes or small task completion directly within the product.

The goal is to provide a cohesive experience where the agent understands the product's current and future states, making it an effective tool for automating development processes and integrating deeply with how teams already work within Linear.

> One of the problems I see, like when I use this, like Claude or ChatGPT or some of these tools or Codex, is that like I have to really explicitly tell the agent or like the, the tool always like what, like, what context to bring.

## 34:04 - 36:05 Linear's AI agent integration leads to usage-based billing for intensive features

Linear, a SaaS company, made a strategic decision to integrate its own AI agents directly into its platform. Previously, the company benefited from being a control hub for other coding agents without incurring AI token costs. The shift to an in-house agent model is driven by the desire for tighter integration and the ability to offer more powerful, unique features to users.

This integration significantly alters Linear's margin profile by introducing substantial token costs. The company recognized that while a direct integration offers product advantages, it requires a careful re-evaluation of its business model to account for these new expenses.

For high-intensity AI features, specifically coding agents, Linear plans to implement a usage-based billing model. This approach is necessary because the operational costs associated with these features can become very expensive, making it unsustainable to include them within a standard fixed-price subscription.

Conversely, basic AI functionalities, such as agents that answer questions or provide general assistance, are expected to remain included within the existing Linear subscription. These features are less resource-intensive and can be absorbed without additional charges, maintaining value for all users.

> On the coding agents, like, we, we do have to offer like usage based billing because it can get very expensive.

## 36:05 - 38:05 Linear aims to be a focused product context and memory platform

Linear is not designed to be a generic agent platform where users run random tasks. The platform maintains a clear focus on specific workflows and workloads, ensuring its purpose remains distinct and well-defined for its users.

Instead, Linear positions itself as a "Product Context or Product Memory platform." It is built to provide a central place for product-related information and insights, allowing for integration with various agents.

This approach enables Linear to act as an API into product thinking. It contrasts with typical tools that require users to constantly fetch information because they lack inherent understanding of the user's general activities or existing context.

Upcoming features for Linear include "skills," which will offer both organizational and personal guidance, allowing users to define and utilize specific capabilities within the platform. They plan to have classic input boxes and contextual interfaces based on projects.

> it's just more like the Product Context or the Product Memory platform where you can integrate those agents... it's just a way to work around your product and it's kind of like an API into the product thinking.

## 38:05 - 40:05 Linear demonstrates AI agent skills for product problem synthesis and direction.

Karri Saarinen showcases Linear's custom AI agent skills designed to assist as a product teammate. These agents are trained using internal company materials, such as blog posts, to help product teams quickly understand complex feature requests and user problems.

The agents operate with a specific structure, first identifying the underlying customer need before diving into the details of the problem. For instance, when analyzing the request for "multiple workspaces," the agent processes various customer inputs and activities to synthesize the core motivations.

The AI agent found that companies frequently seek multiple workspaces to manage different internal divisions while maintaining a single, unified point for billing and governance. This demonstrates how the AI tool provides structured insights to guide product development. Linear currently uses the Claude model for these agents.

> act like a linear product teammate and, and then it, it has this format of like, it starts with the underlying need and it has this like way it goes through the problem

## 40:05 - 44:06 Linear's AI Coding Agent Facilitates Collaborative Development and Code Review

Linear demonstrates how its AI coding agent can be tasked to create new features, such as a new dark theme. The agent integrates this request by turning it into an issue, delegating it within the system, and then beginning its work in a sandbox environment.

A core benefit is the shared context of the agent's work session, which is visible to the entire team. This allows colleagues to observe the agent's progress, jump into a shared chat to tweak the work together, or see what changes have been made. This transparency helps clarify the origin of changes, whether from a user or a customer discussion.

The shared context proves valuable for cross-functional collaboration. For instance, a product manager and a designer could iterate on UI tweaks, both seeing real-time preview links and code diffs. This setup collapses the traditional feedback loop by allowing immediate interaction with the agent.

This workflow extends effectively to code reviews. Instead of engineers verbally communicating changes, they can directly ask the agent to implement fixes. This mechanism streamlines the review process, enhancing efficiency and enabling multiple team members to collaboratively guide the agent on a single task.

> It kind of collapses the collaboration loop a lot more, and allows multiple people to use the agents to work on one thing.

## 44:06 - 46:06 Linear's Strategic Focus on Upstream Workflow Automation

Linear is expanding its product capabilities, which prompts questions about increasing its surface area and competing with existing solutions, especially in the rapidly evolving field of AI coding. The challenge lies in potentially recreating features that are already well-established by other companies.

Karri Saarinen clarifies that Linear's unique advantage is found in positioning itself "upstream where the work is coming from." This strategic placement allows them to offer significant leverage by automating the initial stages of work.

For instance, when new work items or bugs are reported, they can be automatically assigned to agents and delegated, often presenting engineers with tasks where a fix is already being developed. This approach allows engineers to focus on higher-value work.

Linear is not attempting to solve every coding need or create AI agents capable of building entire new products. Instead, their focus is on optimizing workflows for large companies by efficiently handling routine, predictable tasks.

> I think it's, it's what we see the value is is kind of like sitting upstream where the work is coming from. There's like really like good leverage there.

## 46:06 - 48:06 Linear optimizes product workflows to help companies output better and faster.

Linear's product philosophy avoids becoming a "kitchen sink" solution that tries to do everything for everyone. This broad approach often results from enterprise buyer checklists, leading to products that lack a cohesive user experience.

Instead, Linear identifies and optimizes the natural next steps within specific product development workflows. For instance, if an issue is reported, the immediate next step is to fix it. The company focuses on streamlining this process, potentially using tools like AI coding agents, to enable quicker resolutions.

The core goal is to help companies produce better and faster outputs without attempting to own every single surface or aspect of the development process. By concentrating on these optimized workflows, Linear enhances efficiency for particular product tasks rather than expanding broadly.

> we are kind of like trying to find this optimized workflow for people to do certain kind of like product things.

## 48:06 - 50:07 Envisioning Self-Driving Agents in Product Development

Karri Saarinen discusses a future for product development that includes "self-driving" capabilities. This involves establishing clear rules or guidance and leveraging a "project memory" or common workflows to manage incoming feedback and requests.

He envisions products or features behaving like agents that can make autonomous decisions based on input and identified patterns. These agents could propose potential solutions, build and test features with customers, and incorporate their feedback, effectively automating parts of the development cycle.

Even with these automated agents, humans remain essential. People will need to be more explicit about their desired outcomes, what features are worth pursuing, and which areas to focus on. Human involvement in meetings, discussions, and documentation will still be critical for understanding and guiding the work.

> you can't just outsource the thinking purely

## 50:06 - 52:07 Human intuition and clarity remain central to product development despite AI advances

Despite the rise of AI agents, human roles in product development are expected to shift, not be entirely replaced. Karri emphasizes that clarifying one's own thinking and strategy is crucial, not only for human teams but also for codifying instructions for autonomous AI systems. This suggests that human strategic input becomes even more vital in an AI-assisted future.

Product building is viewed as a craft or an art, heavily reliant on human intuition. Decisions are often made based on an inherent understanding of the problem, with data primarily serving as a signal rather than the sole driver. This approach contrasts with purely data-driven methods like A/B testing, which might be suitable for AI agents but not necessarily for all types of products.

The best products are not always built by rigid data-driven processes. They still require a distinct human touch to determine what makes something truly interesting, valuable, or "good." This human element, involving qualitative judgment and creative insight, is seen as irreplaceable by AI.

> I still don't see like how the agents, like, or how the AI actually like does all the thinking and like, kind of like the choices or decisions. I think product building is still kind of like a craft or an art.

---

Get podcast briefs for shows you follow: https://podbrew.app/