Ask your AI assistant about your feedback.
InstantFeedback runs an MCP server, so an AI assistant can explore your feedback the way an analyst would — which campaign slipped this month, what the detractors keep mentioning — working from exact numbers, read-only, and without seeing anyone's personal data.
- Connects with
- A remote MCP server, signed in with OAuth
- Access
- Read-only — ten tools, none of which can change anything
- Can explore
- Metrics, feedback, comments, tags, campaigns and surveys
- Controlled
- Per user, per AI client and per campaign
An analyst's access, with the guard rails built in.
Ask in plain language
Put questions to your feedback in your own words and let the assistant work through it: find the lowest grades, read the comments behind them, and follow the thread across campaigns and tags.
Exact numbers, not estimates
The assistant doesn't do the maths. InstantFeedback computes CSAT, NPS, average grade and response, comment and survey completion rates itself, split by campaign, channel, tag or your own custom fields, or by day, week or month.
Surveys, question by question
Survey funnels and drop-off, the results for each question, and a sample of the written answers — so the assistant can tell you where people give up and what they say before they do.
Read-only by design
Every tool the server offers only reads. An assistant can explore feedback, tags, campaigns and surveys, but it can't send a request, change a setting or delete a thing.
Personal data stays out
Feedback reaches the assistant without phone numbers or email addresses, without any field or survey question flagged as personal data, and without anything that's already been masked.
Access you decide
Switch it on user by user. Choose whether new AI clients are allowed or blocked by default, let people request access to a blocked one, and keep any campaign out of reach entirely.
Three steps, then just ask.
Turn it on
Activate the MCP server for the account, then switch it on for the users who should have it — it starts off for everyone.
Connect the assistant
Add InstantFeedback to the AI assistant as a remote MCP server and sign in. InstantFeedback asks you to approve the connection before anything is shared.
Decide what's in scope
Set whether new AI clients are allowed by default, answer access requests, and exclude any campaign that should stay out.
Before you ask.
Which AI assistants can connect?
Assistants that support remote MCP servers and sign in with OAuth using a client ID metadata document. We'll confirm yours works as part of the demo.
Can the assistant see who said what?
It sees the grades, comments and survey answers, but not phone numbers, email addresses, fields or questions you've flagged as personal data, or anything already masked.
Can it change anything?
No. The server is read-only: there's no tool that sends, edits or deletes anything in InstantFeedback.
Can we keep some campaigns out of it?
Yes. Any campaign can be excluded, and its feedback and results are then invisible to every AI assistant, for every user.
Which plan includes it?
Integrations are part of the Business and Enterprise tiers. See pricing for what each tier includes.
The rest of your stack.
HubSpot
Fire feedback requests from workflows, tickets and deals. Grades, comments and survey results land back on the contact, deal and ticket timelines.
How it worksEmarsys
Trigger feedback requests from Emarsys programmes, and send grades, comments and survey results back out for Emvents to carry into Emarsys.
How it worksSlack
New grades and comments posted into the channels your team reads, with comments routed by grade and daily, weekly or monthly reports alongside.
How it worksMicrosoft Teams
New grades and comments posted into Teams channels as cards, with comments routed by grade and daily, weekly or monthly reports — connected through a Workflows webhook.
How it worksWebhooks
Six events — request sent, reminder sent, grade, comment, survey started, survey completed — pushed to any endpoint, HMAC-SHA256 signed.
How it worksSee it answer questions about your own feedback.
On a call we'll connect an assistant to a demo account and put it through the questions your team actually asks.