AI analysis overview
The AI Analysis feature helps admins quickly understand, categorize, and analyze large volumes of resident input. It works for both ideation and surveys, offering:
Qualitative analysis – Summarizing textual input from ideas, comments, or survey responses.
Quantitative analysis – Exploring correlations between demographics, tags, and survey responses.
Access depends on your plan (see “Price Plan Capabilities”).
AI analysis interface
The interface has four main columns:
Tag Management: Create and manage tags for clustering inputs.
Input List: Shows all contributions to analyze.
User & Filter View: View anonymized user responses, apply filters, or drill into comments.
Summaries & Questions: Summarize filtered inputs or ask AI for insights.
How to use AI analysis for ideation/proposals?
Go to the project’s Input Manager.
Launch AI Analysis.
Summarize inputs and comments (scroll to the comment section in column 3).
Use filters (e.g., by demographics) to narrow focus before summarizing.
How to use AI analysis for surveys?
Go to the Survey Results page.
Click the blue Open AI Analysis widget.
Select questions you want to analyze.
Optionally, add follow-up questions for deeper analysis.
How to filter and preview inputs?
Use filters by timeframe, engagement level, or demographics.
Preview how answers distribute across demographics (only for users who completed their profile).
How to use summarize and ask a question?
Use Summarize to create a digest of selected inputs.
Use Ask a Question to probe further.
Check in-line references and validate with original inputs.
Drag summaries into reports via the Report Builder (summaries appear in the “AI” tab).
Tagging methods in AI analysis
Tagging helps cluster and interpret user input. The tool supports:
Fully automated tagging: AI assigns common tags.
By label: Create your own tags.
By example: Provide manual examples to teach AI tagging.
Sentiment tagging: Mark input as positive or negative.
Language detection: Identify the input language.
Manual tagging: For full control, you can tag every input manually.
What are auto insights?
Auto‑Insights connects tags, demographics, and survey questions to uncover patterns.
Displays key correlations (e.g., which age group engages with which tag).
Includes a heatmap view, with boxes marked if a correlation is statistically significant.
Unit of analysis can be switched between:
Inputs (contributions)
Likes
Dislikes
Participants
Difference:
In ideation, Auto‑Insights connects demographics to tags.
In surveys, Auto‑Insights links demographics, tags, and survey questions.
You can get into the auto-insights by pressing 'explore' below the demographic information of the AI tool interface.
Why is human oversight recommended?
The Summarize and Ask a Question tools generate quick insights but require human oversight.
Summaries include clickable references to original inputs for transparency.
Overloading the AI with too many inputs can reduce accuracy—filters & tags help.
Missing in-line references may happen for reasons like:
Contextual relevance (references not needed)
Balancing readability (avoiding excessive citations)
System refinements (constantly improving).
Note: Summaries are not 100% accurate. Always cross-check key findings with the original data.
External providers and data processing
The AI Sensemaking tool uses Microsoft Azure OpenAI APIs (GPT‑4 Turbo) for summarization, tagging, and question features.
Microsoft processes data only to provide and support the service.
Data is processed in one of 4 customer regions (EU, UK, US, CA).
Data does not go to OpenAI’s public API or ChatGPT.
Only text written by users is sent—no usernames, emails, or PII unless a user writes it into their input.
Data is not used to train the models.
Data privacy
Only text written by residents is sent.
No usernames, emails, or demographics sent unless mentioned in the text itself.
Data is processed in your selected region (EU, UK, US, or CA).
Providers (Microsoft Azure) do not use your data to train their models.
AI analysis FAQ
What’s the maximum amount of inputs the AI can process at once?
AI can process ~400,000 words at once (depends on language, content length, number of questions).
If you see the message “AI can’t process so many inputs in one go,” split inputs into smaller groups.
For what pricing plan is AI analysis available?
Essential: No AI summaries.
Standard: AI summaries up to 30 inputs at a time.
Premium & Enterprise: Unlimited AI summaries.
Is there support for different languages?
AI responds in the language of the input whenever possible.
Mixed-language or low-data situations may cause wrong-language responses, retrying usually fixes this.
All core platform languages are supported (except Greenlandic).