HAI is the AI layer built into Harvestr. It works in the background to help your team process incoming feedback faster and surface insights without manual effort.

Four core capabilities:

HAI is included on every paid plan. There's no AI add-on to unlock.

What differs between plans is your monthly allowance of AI categorizations. One categorization covers one feedback item — one support ticket, one call transcript — regardless of how many verbatims are extracted from it.

HAI processes incoming feedback automatically, extracts the most relevant verbatims, and suggests which Discovery to link them to. You validate or reject those suggestions, individually or in bulk.

Whether it's support tickets, sales calls, or Slack threads, HAI scans your feedback and highlights what's worth keeping.

Suggestions are marked in purple inside the feedback content, and listed in the Suggestions section at the bottom of each item.

At the top of each item you'll see how many suggestions were found. Click it to jump straight to them.

This quick filter shows every item in your inbox that currently has suggestions:

After extracting verbatims, HAI suggests where they belong:

For each suggestion you have three options:

All of this works individually or in bulk.

HAI handles the first pass, but you can always categorize by hand — and you'll want to for anything it missed, or for feedback you've already marked as processed.

Click Link to at the top of the item.

A popup shows all your Components and Discoveries. Click one to link the feedback to it, or use the + button next to a Component to create a new Discovery on the spot.

Often only some of what you receive is relevant, or a single item covers several parts of your product.

Select the passage you want with your mouse and the same popup appears, this time linking only the extract you highlighted.

You can keep extracting several passages from the same item, each to a different Discovery, then mark it as processed when you're done.

You can also work from the other end. In the Discoveries view, the feedback column shows how many suggestions HAI has identified for each Discovery across all unprocessed feedback — patterns surface without combing through your inbox.

Inside a Discovery, the Suggestions tab lists every related verbatim, ready to validate or reject.

Long call transcripts and user test reports can be summarized in one click, pulling out the valuable parts without reading the whole thing.

A well-used Discovery accumulates dozens of verbatims. HAI summarizes the whole set, so you can identify patterns and trends rather than re-reading everything each time you revisit the topic.

Set the language HAI writes summaries in from your personal settings.

English is the default. Also available: French, Spanish, German, Italian, Danish, Dutch, Georgian, Greek, Norwegian, Czech, Portuguese, Arabic and Swedish.

This is a personal setting — teammates can each use a different summary language.

No. HAI is included on every paid plan, with the monthly categorization allowance listed above. See the pricing page for details.

Once enabled, every Editor in your workspace has access.

HAI works with a wide range of languages, though it's tested and developed primarily in English.

The latest models from Anthropic and OpenAI. Using HAI means accepting Anthropic and OpenAI as subprocessors of your data.

Yes. All our security policies are in our Trust Center.

Your data is not used to train any model. OpenAI may securely retain API inputs and outputs for up to 30 days for abuse detection, then deletes them. More on OpenAI's enterprise privacy.