Granola's $1.5B valuation isn't about being a better note-taking app

Granola closed a $125 million Series C in March 2026 at a $1.5 billion valuation, up from roughly $250 million a year earlier, after revenue grew 250% in a single quarter. For a note-taking app that started as a way to clean up meeting notes, that's an odd number to defend on note-taking alone.

It makes more sense once you look at what the company shipped alongside the round: Spaces, a team workspace with folder-level access controls, and two new APIs. One lets a person pull their own notes, and notes shared with them, into other software. The other lets an admin expose a team's notes as context for AI workflows. Granola didn't raise to take better notes. It raised to become the thing other software reads from, the same way a database becomes infrastructure the moment something other than a human starts querying it.

Three jobs hiding inside every note-taking app

"Note-taking app" is a single phrase covering three genuinely different jobs, and most of the friction people feel with these tools comes from using a tool built for one job to do another.

The first job is capture: get the thought down before it's gone. Apple Notes and Bear are built almost entirely around that single goal. Open fast, type, close, with no required structure before the text counts as saved. The second job is connection: make a note findable and linkable months later, which is what Obsidian, Logseq, and Notion-as-wiki optimise for, usually at the cost of capture speed. The third job is newer: active workspace, where the note is useful the moment it's written because it's plugged into a live piece of work happening right now.

Granola, and tools like Storyflow and Mem, sit in that third category. The note isn't an artifact to file away. It's an input something else acts on immediately, often before the meeting that produced it has even ended. That's a different design target than either capture or connection, and it explains why a tool built for one job tends to feel clumsy the moment a team tries to make it do another.

Why teams keep switching and never settle

Knowledge workers reportedly switch between apps and tabs over a thousand times a day, and lose close to a tenth of their working time to that switching alone. Roughly 60% of the working week now goes to what gets called "work about work": searching for information, moving between tools, chasing down decisions that live in someone else's app. None of that is a notes problem specifically, but notes sit right in the middle of it, because a note is usually the thing someone is searching for when the search fails.

This is the part most "best note-taking app of 2026" roundups skip. They compare apps on a shared feature checklist: tags, backlinks, mobile sync, AI summaries, as if every team buying a note-taking app is solving the same problem. They aren't. A support team capturing call notes and an engineering team building a shared design doc are using the word "notes" to describe two different jobs, and a feature checklist can't tell you which one you have.

The tell that you picked the wrong job

The pattern shows up the same way almost every time. A team adopts a connection-first tool, usually because someone senior already uses it personally and likes its structure. Within a few weeks, daily usage drops, because connection-first tools ask for a small amount of upfront organisation, such as a folder, a tag, or a linked page, before a note counts as saved. That's a fine trade for a personal knowledge base built up over years. It's friction for a team trying to capture what just happened in a stand-up, where the value of the note is inversely related to how long it took to write.

The reverse mismatch is just as common, and quieter. An engineering team starts a shared runbook in a capture-first tool because it was already on everyone's laptop and felt like the path of least resistance. Six months in, the runbook has forty entries, no links between related incidents, and nobody can find the one note that explains why a particular service was reconfigured last spring. The tool did exactly what it was built to do. Capturing was never the problem; finding it again was, and that's a job the tool was never asked to do well.

The fix usually isn't a better tool. It's noticing that the team had one job and bought a tool for a different one. Sometimes that means swapping the tool. Sometimes it means accepting that two separate note-taking tools for two separate jobs isn't redundancy. It's just correctness.

What happens once the note becomes AI context

Granola's enterprise API is the clearest signal yet that the active-workspace category is becoming infrastructure rather than an app people open on purpose. Once notes are exposed as context an AI agent can read and act on, the interesting question stops being "how do I find this note again" and becomes "what is this agent allowed to infer from it, and on whose authority."

That's why Spaces ships with folder-level access controls rather than a simple shared-or-private toggle. A note that only a human will ever read is low-stakes if the wrong colleague sees it. A note that an agent reads before taking an action, such as drafting a reply, updating a record, or scheduling a follow-up, carries the access-control weight of the action it might trigger, not just the weight of the text itself. Teams adopting active-workspace tools for that reason should expect to spend real time deciding who can grant an agent access to which folder, which is a different conversation than who can edit a shared doc.

A three-question diagnostic before you adopt or drop a tool

Skip the feature comparison and ask three things instead.

• Where does this note need to live ten minutes from now? If the answer is "saved, doesn't matter where," you have a capture job.
• Where does this note need to live ten months from now? If the answer involves finding it through a different note that links to it, you have a connection job.
• Does this note feed directly into something else that happens automatically, such as a summary, a follow-up, or an agent's next action? If yes, you have an active-workspace job, and the questions that matter shift from search to access control.

Most tool churn disappears once a team can answer those three questions separately for each kind of note they take, rather than picking one app and hoping it covers all three. A single team will often have all three jobs running at once: capture in standup, connection in a wiki, active workspace in whatever now plugs into an agent. That's not a failure to standardise. It's three different jobs that happen to share the word "notes."

The teams getting real value out of this category in 2026 aren't the ones with the most polished notes. They're the ones who stopped expecting one tool to do all three jobs at once, and started budgeting for the access-control conversation before an agent, not a colleague, became the thing reading what they wrote.