Your AI companion for Notion.

Written down is not the same as findable.

You know the team documented the onboarding process.

You do not know whether the live version is the page in the handbook, the one in the old project folder, or the one somebody titled “Notes v2”.

Searching by title does not settle it, because the useful page is rarely named after the thing you are looking for.

A companion can open the candidates and read them properly. It can pull the block contents of each page, tell you what each one actually says, and show you where they contradict each other. It can also tell you which was edited most recently.

That turns a wiki archaeology afternoon into a short answer with links, and it gives you the list of pages worth archiving.

A tracker nobody updates is worse than no tracker.

People trust a half-filled database anyway.

The vendor comparison was accurate in March. The status column says a project is in progress and it shipped weeks ago.

The reason those cells stay empty or wrong is usually that the real value lives somewhere else. It is on forty vendor websites, in a Gmail thread from six weeks ago, in a Linear issue, or on a pricing page nobody wants to open forty times.

Strawberry can go and get it. It can query a database, take each row’s subject back out to the live web or to another connected tool, and prepare the updated properties for your review. Gmail, Outlook, Slack, Linear, Jira, GitHub, Google Sheets and Google Drive are native integrations, so context from those tools can feed the same pass.

Ask it to flag rather than guess. The vendor whose pricing page says “contact us” then comes back marked for a decision, not filled in with a number that looks right.

Notes only become work if someone types them in again.

The call happened, the decisions were real, and the notes are sitting in a doc in the shape a human takes notes. Somebody now has to reread them and turn them into a page, a set of rows, and an owner per item. That job lands at the end of the day, which is why it often does not happen.

Strawberry can do the shaping. It can create a page under the parent you name and append the content as proper blocks. It can insert the action items as rows in the right database, with an owner on each, and leave a comment where something is still unclear.

The same applies to a database you do not have yet. Describe what you want tracked and a companion can create it with properties that match the question, rather than a default Name and Tags. It can fill the first rows too, with a source link on each.

The follow-up should start when the page changes, not when you remember.

A request gets filed in a database and sits there for two days, because the person who files requests is not the person who reads them. A comment asks a question on a spec and gets answered in a meeting instead, so the page still shows an open question a month later.

Strawberry can start work from a change in Notion. A routine can run when a page is added to a database, when a page is created or updated, or when a comment appears. It then does the next step rather than notifying you about it.

A companion can read the comments on a page and reply in the thread with your approval. The same routine can run on a plain schedule instead, if a weekly pass suits the work better. See how routines work.

Editing a wiki is not like sending an email.

A bad email is embarrassing and obvious.

A page that quietly changed under twelve people is neither, which is why write access to a workspace deserves more care than most connections.

Permission settings control which actions require your approval, so you can read the proposed edit before it reaches the page. Archiving a page or a row moves it to Notion’s trash and it can be restored. Removing a property from a database is different, because the column and everything in it goes with it. Keep approval on schema changes, as set out on the security page.

Frequently asked questions