If your sales and customer success teams use Praiz to record and transcribe calls, you can pipe the product feedback those calls contain into Harvestr automatically. A dedicated Praiz AI agent identifies feedback in each call, structures it into fields, and sends only the relevant calls to your Harvestr inbox — with the customer and call metadata already attached.
Prerequisites
An Editor account in Harvestr.
A Scale subscription (or higher) in Harvestr.
A Praiz account with permission to manage AI agents and webhooks.
How it works
Praiz applies an AI agent called Product Feedback to every call you choose to run it on. The agent checks whether the call contains product feedback, and if it does, extracts that feedback into structured fields. Praiz then fires a webhook that sends the call and its extracted feedback to Harvestr.
The result: only calls that actually contain product feedback land in your Harvestr inbox — no manual triage, no noise from calls that aren't relevant.
Each new feedback in Harvestr arrives with:
the linked customer (from the Praiz call attendee)
call metadata (date, participants, and so on)
the content of the Product Feedback agent's fields, which become the body of the feedback
From there, Harvestr's own AI categorizes the feedback into the right discoveries.
Set up the integration
Step 1: Import the Product Feedback AI agent in Praiz
In Praiz, go to AI agents.
Browse the agent library.
Find the Product Feedback agent and import it as public into your workspace.
Step 2: Customize the agent (carefully)
The default agent works out of the box, but you can tailor it to the kind of feedback your team cares about. You can:
Add fields for any structured insights you want to capture.
Remove fields you don't need.
Edit the prompt or description on the fields you keep.
Two things you must not change
The agent title — Harvestr identifies the right Praiz agent by its title. Renaming it will break the integration.
The Product feedback mentioned field — this Boolean field tells Harvestr whether to import the call. If you change or remove it, calls without real feedback will flood your inbox (or no calls will be sent at all).
Quality tip: the values produced by your customized fields become the body of the feedback in Harvestr, and Harvestr's AI categorizes them from there. Keep noise out — focus the fields on the actual product insights you want surfaced. Less is more.
Step 3: Auto-apply the agent to the right calls
In Praiz, open the automation rules for the Product Feedback agent.
Configure the rules so the agent runs automatically on the calls you want analyzed — for example, all customer calls, all calls from a specific team, or all calls tagged a certain way.
Auto-applying the agent is required for the webhook to fire. Without an automation rule, the agent won't run on incoming calls and nothing will be sent to Harvestr.
Step 4: Connect Praiz to Harvestr via webhook
In Harvestr, go to Settings → Integrations → Praiz.
Copy the webhook URL.
In Praiz, go to Settings → Webhooks.
Click Add webhook, paste the URL, and set the webhook type to Agent.
Save the webhook.
What happens next
From this point on, every time the Product Feedback agent runs on a call and detects product feedback:
Praiz fires the webhook to Harvestr.
Harvestr creates a new feedback in your inbox with the customer linked, the call metadata attached, and the content of the agent's fields as the feedback body.
Harvestr's AI categorizes the feedback into the most relevant discoveries.
Calls that don't contain product feedback never reach Harvestr — they're filtered out on the Praiz side.