Skip to main content

HAI - the Harvestr AI

AI-powered features to help you process feedback more efficiently and uncover valuable customer insights faster than ever

Marina Salmon avatar
Written by Marina Salmon
Updated this week

What is HAI?

HAI is Harvestr's AI agent, built to help businesses analyze customer feedback at scale.

It saves your team hours of manual work while uncovering valuable insights to drive product discovery and prioritization. HAI currently offers four core capabilities:

  • Extracting relevant verbatims from all incoming feedback

  • Automatically categorizing feedback by linking it to Discoveries

  • Surfacing insights from large volumes of feedback related to a specific Discovery

  • Summarizing individual feedback notes

Prerequisites

HAI is available on our Scale and Elite plans.

Extracting and categorizing feedback from your Inbox

HAI automatically processes all your incoming feedback, extracts the most relevant verbatims, and suggests the appropriate Discovery to link them to. You can validate or reject these suggestions, either individually or in bulk.

Extracting valuable verbatims from all your feedback

Whether it's support tickets, sales calls, or Slack threads, HAI scans all your feedback and highlights the most valuable verbatims.

Suggestions are marked in purple within the feedback content and are also listed in the Suggestions section at the bottom of each item in your Harvestr inbox.

HAI only scans uncategorized feedback in the New folder of your Inbox. Once feedback is marked as processed, its Suggestions are deleted and it will no longer be scanned anymore.

At the top of each feedback item, you’ll see how many Suggestions were extracted (if any). Clicking this will take you straight to the Suggestions section, where you can review and act on them.

You can also use this quick filter to view all feedback in your Inbox that currently contains suggestions:

Pro tip: For support tickets from Zendesk and Intercom, you can choose to scan only internal notes instead of the entire conversation. If your support team summarizes the feedback in those notes, it helps the AI focus on the most valuable parts. Contact us to active this option.

Automatically categorizing feedback

After extracting verbatims, HAI will suggest the most relevant Discovery to categorize the feedback.

  • If a suitable Discovery already exists, HAI will recommend linking the feedback to it.

  • If not, it will suggest creating a new Discovery, with a proposed title based on the feedback and your existing naming patterns.

For each Suggestion, you have three options:

  • Accept it

    • If it points to an existing Discovery, the feedback will be linked automatically.

    • For a new Discovery, you’ll be able to review and edit the title, description, state, tags, and assignee before it’s created.

  • Reject it

    • The Suggestion will be deleted.

  • Link it to another Discovery

    • The extracted verbatim remains, but you can manually choose a different Discovery (existing or new) to link it to.

All AI suggestions can be validated or rejected individually or in bulk.

Surfacing Suggestions from your Discoveries

You can also track HAI-identified feedback directly from the Discoveries view.

In the feedback column, you’ll see how many Suggestions have been identified for each Discovery across all unprocessed feedback. This helps surface patterns without combing through your inbox.

Inside a specific Discovery, the Suggestions tab shows all related verbatims. You can validate or reject them from here as well, either one by one or in bulk.

Summarize lengthy feedback

With HAI, you can summarize all this lengthy feedback notes and extract the most valuable insights in just a click, saving you hours per week on feedback management and product discovery work.

Extract insights from all the feedback in a Discovery

Discoveries can contain high volumes of feedback and customer verbatims that take time to analyze. With HAI, you can easily summarize all this feedback, saving you hours of manual work and helping you identify patterns and important trends for specific Discoveries. This not only saves you time and effort but also ensures that you are able to focus on the most important customer problems.

Configure the language of AI summaries

You can select the language that should be used by the AI to write your feedback summaries.

This can be set in your personal settings.

English is the default language, and the other languages currently available are: French, Spanish, German, Italian, Danish, Dutch, Georgian, Greek, Norwegian, Czech, Portuguese, Arabic, and Swedish.

This setting will be applied only to you, which means that different teammates in your organization can have different summary languages.

FAQs

Are there any additional costs to using these features?

The HAI features are available on our Scale and Elite plans with usage limits. For additional information, visit our pricing page.

If I enable these features, will my whole team get this or only me?

Once enabled, all the Editors of your workspace will get access.

What languages do these features work with?

HAI works with a wide variety of languages, but it is being tested and developed in English.

Which AI technologies does HAI rely on?

We use Anthropic’s models via AWS Bedrock. Since Harvestr is also hosted on AWS, your data remains within the same provider, and Anthropic does not act as a sub-processor.

Feedback summaries leverages OpenAI’s models and tools. If you choose to use feedback summaries, you agree to OpenAI being a subprocessor of your data.

Is my data safe?

Yes. You can review all our security policies in our Trust Center.

Your data won't be used by OpenAI to train its own models. OpenAI may securely retain API inputs and outputs for up to 30 days to identify abuse, but it will be deleted afterward. You can find more information about OpenAI’s data security, privacy, and compliance here.

Did this answer your question?