Camaleonic
Survey Intelligence
Iris creates the survey, refines it with you and reads the results: crosses, findings and answers with the real numbers.
Figures validated on the server. Data scoped against injections. AI with spending limits.
01 · Create with AI
Describe the goal. Iris delivers the survey.
Write what you want to measure and iterate in chat. When you accept, the survey lands in the builder ready to edit. Try the three scenarios on the panel.
Questions per generation, by plan. Most generators stop near 10.
02 · Refine in the builder
Iris edits with you. You approve every change.
Iris proposes in streaming; you approve every change as a diff, question by question. Play the queue: three proposals, three decisions of yours.
Did you find the talks and the networking useful?
Scale 1-5How useful did you find the talks?
Scale 1-5How useful did you find the networking?
Two topics in one question bias the answer.
03 · Analyze with AI
Ask your data. Iris answers with the real numbers.
Per-question insight, free sentiment and themes, crosses with validated figures and a chat grounded in your data.
812 open answers analyzed
Best speaker lineup in years; worth the trip from Toronto.
Forty minutes queueing at check-in on day one.
The streams worked fine; the time zone from Singapore was rough.
Overall read: positive, with friction at registration.
How would you rate check-in at the venue?
71% score it 8 or higher; the average drops to 6.2 during the morning rush.
Does NPS change by region?
The Americas drop 18 points below Europe: it matches the off-hours streaming slots.
What sets apart the attendees who plan to return?
84% of those returning score the talks 9 or higher; price barely weighs on the decision.
04 · Intelligence in your reports
Every pinned finding travels into the final document.
Pin any AI finding and it travels into the final document, in your brand.
Internal recommendations hide themselves.
05 · Quality and trust
Quality answers. AI with guardrails.
Attention checks as first-class citizens and AI guardrails that are designed, not improvised.
To confirm you are reading carefully, select the option Agree.
Whoever fails is flagged in response quality.
- Prompt-injection fencing at every entry point.
- Every cited figure is validated on the server.
- Spend limits that fail closed, with a kill switch.
- EU-friendly AI operation, with public subprocessors. See subprocessors
Frequently asked
The AI, without fine print.
A usage unit weighted by effort: light actions, like a chat message, consume little; heavy ones, like generating a full survey, consume more. Every plan includes a monthly pool of credits you can extend whenever you want.
Plan credits reset every month. Purchased top-up credits never expire and are only spent after the monthly pool runs out.
Sentiment analysis stays free. You can buy a top-up with a one-off in-app payment, or upgrade. Expensive actions show a cost estimate first, and the usage meter lives in Settings.
Your call: one shared pool for the whole organization or per-member caps, with individual overrides. The admin manages it from Settings, with a per-person usage breakdown and one-off top-ups when needed.
Yes. The personality panel sets Iris's tone for the whole organization, from neutral and executive to warm, and it applies to chats, summaries and reports.
The General view is on every plan. The Customer view arrives with Pro. The Employee and Suppliers views arrive with Spectrum. Enterprise includes everything, with per-view overrides for custom deals.