AI features
Learn what EnaFeedback's AI features do, from sentiment analysis to voice and video transcription and translation, their defaults and how to switch them.
EnaFeedback includes AI features that help you process feedback and content faster: sentiment analysis on visitor comments, transcription of voice and video attachments, and translation of content. They are assistants; your team always decides who handles a ticket and how it is closed.
This article explains what each feature does, its default state and how to turn it on or off.
Required role
AI settings live under System → System settings → AI preferences. System settings are managed by the Owner and Administrator roles. Connecting your own provider key is part of Integrations and belongs to the Owner role; see Bring your own keys.
Features and defaults
| Feature | What it does | Default |
|---|---|---|
| Sentiment analysis | Judges whether a visitor's text is positive, negative or neutral | On |
| Voice transcription | Turns voice attachments into text | On |
| Video transcription | Turns the audio of video attachments into text | On |
| Dynamic content translation (visitor) | Translates pages, surveys and module text into the visitor's language | Off |
| Comment/message translation (admin) | Translates feedback written in other languages for admins | Off |
| Transcript translation | Translates voice and video transcripts | Off |
For translation details, see Translation.
Where sentiment analysis is used
Sentiment analysis powers the Conditional option of the Positive outcome setting. Positive outcome is an optional thank-you block shown to the visitor after they submit. It is set separately for hygiene, visitor feedback and staff feedback.
- Always: the block is always shown.
- Conditional: the block is shown when AI rates the visitor's text as positive.
- Never: the block is not shown.
A tolerance setting decides what happens when sentiment is neutral or unavailable.
Where transcription is used
- Voice and video recordings attached to feedback are transcribed, so you can read the content without playing the recording.
- For videos you upload to content pages, if you do not upload WebVTT captions, captions are generated from the AI transcript where possible.
- Transcription accepts audio up to 20 MB and video up to 50 MB.
Turn a feature on or off
Open AI preferences
Go to System → System settings → AI preferences.
Pick the feature
Flip the switch for the feature you want to turn on or off.
Save
Once saved, the change applies to new content.
Which provider is used
The default provider is the ENA AI Gateway and needs no setup. You can connect your own API key for OpenAI, Google Gemini or Anthropic Claude instead. One provider is active at a time, and there is no automatic failover to another provider.
Daily usage limits
AI calls have built-in daily limits: 20,000 sentiment calls, 5,000 translation calls and 5,000 transcription calls per day. These limits are fixed and cannot be changed. Details: Bring your own keys.
Logging and transparency
Every AI call is listed under System → Records → AI invocations, with provider, model, latency, character and token counts, and status. Prompts and content are not stored in this log.
Limits
- There is no AI summarization feature.
- AI does not route, close or prioritize tickets. Priority comes from the feedback category.
- Anthropic Claude supports text only, so it cannot be used for voice or video transcription.
Related articles
Can't find what you need?
Our support team can help with anything from setup to reporting.
Exports (PDF and Excel)
Export reports and lists to Excel, tickets, hygiene reports and staff summaries to PDF, save charts as PNG, and see how every export is logged.
Translation and languages
Learn about the four visitor languages, automatic AI translation of your content, and how admins read feedback written in another language.