Linear AI: Is it any good?

At MarsBased, Linear has become one of our core tools. We adopted it back in 2021.

We use it to manage all our development work: projects, tasks, priorities, status updates, roadmaps, and day-to-day coordination. But we also use it internally as an async communication tool. We have a Linear team inside Linear where we share internal threads, announcements, decisions, and things we want the team to keep for later. And, on top of that, we use it to track our sales pipeline, our hiring processes and also our editorial content for marketing, too!

Slack and Microsoft Teams are great for informal communication, quick questions, and things that don’t need to be stored forever. Linear is where important work-related communication lives. If something needs context, ownership, visibility, or follow-up, it usually belongs in Linear.

After trying Jira, GitHub Issues, Trello, Monday, Asana, and many other tools over the years, Linear is by far the best task management tool for software development we’ve used. It’s fast, convenient, well designed, and it just works.

So when Linear started adding more and more AI features, I was not immediately convinced. My first reaction was: do we really need another AI layer inside Linear?

At MarsBased we already use Claude Code, Gemini, Codex, and other AI tools depending on the project and the team. Developers already spend a lot of time in coding tools, terminals, IDEs, and AI coding assistants. And with Linear’s MCP server, external AI tools can already interact with Linear quite well.

I wasn’t sure this was the best direction for Linear. There were other features I personally wanted more, like better team-level documents, communication boards, or stronger async collaboration features. I was worried that AI could become a distraction from the core product (like it's happening in many SaaS products). But after trying the latest Linear AI features, I think I was probably looking at it from the wrong angle.

This is not mainly for developers

For developers, Linear AI is useful, but probably not where the biggest impact is.

Developers usually live closer to the code. Their AI workflows often happen inside Claude Code, Codex, or Cursor. Linear is important, but it’s not always where the deepest work happens. For product managers and project managers, it’s different. A lot of their work happens directly in Linear, or ends up there somehow.

• When we define a feature, it becomes Linear issues.
• When we organise a project, we create boards, projects, milestones, and overviews.
• When we prepare a report, a lot of the information comes from Linear.
• When we prioritise work, we do it through Linear.
• When we follow up on decisions, we check issues, comments, statuses, and project updates.

So for PMs, having AI inside Linear is actually a very natural step.

What Linear AI gets right

The most interesting part is that Linear AI understands where you are: If you open it while looking at a board, it has the context of that board. If you open it inside an issue, it understands that issue. If you are reviewing a project, it can work with that project context.

Yes, you can do similar things with Claude Code or Codex using Linear’s MCP server. That’s how we’ve been using it so far, and it works well, but having it directly inside Linear removes a lot of friction.

You don’t need to switch tools. You don’t need to copy and paste context. You don’t need to explain which board, project, or issue you’re talking about. You’re already there.

That makes it especially useful for things like:

• drafting or improving issues
• summarising a project
• reviewing a board
• identifying unclear tasks
• preparing status updates
• asking questions about priorities or blockers
• turning rough notes into structured Linear issues

That’s where I think Linear AI can be genuinely useful.

What I like so far

The first thing I like is that it feels fast. That matters because Linear’s speed is one of the main reasons we like the product. If AI made Linear feel slower or heavier, that would be a problem. So far, it feels quite snappy.

I also like that it understands the current view pretty well. If you are inside a board or an issue, the context is already there. That makes the interaction feel much more natural than opening another AI tool and rebuilding the same context manually.

Linear also has skills, which can be customised. That could be very useful for repeated PM workflows like project reports, issue refinement, triage, or weekly updates.

What is still missing

There are still some things we don’t like, or at least things we haven’t solved yet.

As far as we’ve seen, skills cannot easily be shared between team members. That’s a limitation for us, because we don’t only want individual productivity. We want reusable workflows across the team.

Another limitation is that the chat only lives inside Linear.

That makes sense, of course, but PMs don’t only work in Linear. They also spend a lot of time in Figma, Google Drive, Gmail, Google Meet, Slack, and other tools. With something like Claude, you can invoke it globally with a shortcut while you’re in another app. Something similar for Linear-aware workflows would be very interesting.

I’d also love to see a voice mode. I dictate a lot of things, and being able to turn spoken thoughts into issues, comments, reports, or project notes would be very useful.

We also still need to test how much context it really understands. For example, whether it properly reads comments, understands status changes, and distinguishes old information from current decisions.

My take

I was skeptical about Linear AI at first because I didn’t want Linear to lose focus, but now I think it makes a lot of sense, especially for product managers and project managers.

Developers will probably keep using AI mostly from their coding tools. But PMs spend a lot of time in Linear, and most of their work either starts there, ends there, or depends on information stored there.

So having an AI assistant inside Linear feels like a natural evolution.

If Linear keeps this fast, focused, and useful, I think it can become a very valuable part of how product and project teams work.