All nine hundred posts read and grouped by what they actually ask for, each cluster carrying its combined vote count and the posts inside it.
Nine hundred posts across four boards. The same request appears five times under five titles because five people searched with different words, and twelve posts have said “planned” since December.
A companion reads the posts rather than counting them: the details and comments of each, grouped into clusters, with the “bulk export”, the “download all my data” and the “CSV of everything” named as one feature.
01
What is actually being asked for, across 900 posts?
A companion works through the board with the details and comments of each post and groups the requests that describe the same need under different words. Each cluster comes back with its combined vote count and the posts inside it.
Votes are part of that picture, listed per post and per user, which turns “this has 40 votes” into “this has 40 votes and eleven are from accounts on your enterprise plan”. Assembling that is the data extraction pattern.
02
The requests customers never posted
A support conversation, an email reply, a sentence in a renewal call.
A companion reads where those land and files the post on the right board on behalf of the user who asked, because the create tool takes an author and users can be looked up or provisioned by email or external ID.
Attribution is the point. A post filed under a customer’s name carries their vote and lets you close the loop when it ships. Moving work between systems without a person retyping it is the operations motion.
03
The posts that have said “planned” since December
A companion lists posts by status, pulls the ones that have not changed in months, checks each against your tracker (Linear, Jira and GitLab are native integrations) and reports which shipped, which are in flight, and which died without anyone closing them.
Then it proposes the status changes. Approval matters here: a status change is visible to everyone subscribed and can send notification emails, so it is customer communication with a person behind it.
04
How does this fit with the board you already run?
The companion takes the reading pass: clustering the nine hundred posts that exist, correlating votes with account value from the CRM tab you are signed into, and filing the requests customers made somewhere else.
Canny’s own features, as of this writing, catch a duplicate in the box the customer is typing into and inherit the board’s notification and permission behaviour. Reading the nine hundred posts already on the board is the pass above.
Yes. Canny is a native integration connected with an API key. A companion reads boards, categories, posts, comments, users and votes, creates boards and posts on behalf of a user, updates posts, changes statuses, comments, votes and deletes. Public-facing writes wait for your approval.
It identifies them. A companion reads post details and comments across a board, groups the requests that describe the same need in different words, and reports each cluster with its combined votes. The merge itself is done in Canny by a person.
Yes. Canny’s create tools take an author, and users can be looked up or provisioned by email or your own external user ID, so a request made in email or on a call is filed under their name and carries their vote. You approve before it goes public.
Ask for posts filtered by status that have not moved in months, then have the companion check each against your tracker and propose an update. Status changes are visible to subscribers and can trigger notification emails, so each one is approved individually.
Canny, Productboard, Linear and Jira are all native integrations, so one companion reads the board, clusters the requests, and files the work where you plan it. Download Strawberry free; paid plans start at $20 a month.
Canny Strawberry