These docs describe the current Dante AI platform. On the earlier platform? Your API reference is api-v2.dante-ai.com/openapi.json, and your key goes in the x-api-key header.
Analytics
Conversation volume, assumed resolution, median response time, and your most-asked questions over the last 7, 14, 30, or 90 days.
Where to find it
Insights are per agent. Open the agent you want to review, then select Insights from the left menu. From your dashboard, Analytics opens insights for your first agent.
Available on Starter Advanced Pro. Free workspaces see sample numbers with an upgrade prompt instead of live data.
While the setup wizard is still running, Insights is dimmed in the left menu. The line under it says why and gives you the one thing that opens it: on Free that is a link to your plans, and on a paid plan it is a link to Go live, because the report fills once people start chatting.
Time range
Pick a window of 7, 14, 30, or 90 days (30 by default, the same as your dashboard), or set a custom start and end date. Every metric also shows how it changed against the previous, equal-length period, so you can see whether things are trending up or down. If the previous period had no data at all, the change shows as "New" rather than a 0.0% change. If there were no conversations in the window, assumed resolution and response time show a dash instead of a number, because there is nothing to measure yet.
Metrics
- Conversations, the number of chat sessions started in the window, shown with a sparkline of daily volume.
- Assumed resolution, the share of conversations that ended without a handover to a human. It is an assumption, not a check: a visitor who left without asking for more counts, and it does not verify that the answer was right. On a plan without human handover, or while handover is switched off for your agents, there is nothing to compare against, so it shows as not measured instead of a flat 100%.
- Median response time, the median gap between a visitor message and your agent's reply, in seconds, with the number of replies it is based on. Time a visitor spends filling in a form your agent showed them (for example a lead form) is not counted, because that is the visitor's time, not your agent's. Below 5 replies it says "Not enough conversations yet" instead of a number, because a median of one or two replies is noise.
Top questions
The five questions visitors asked most in the period, shown exactly as they were written. Questions that differ only in capitalisation or punctuation count together, each is counted once per conversation, and your own test chats in the Sandbox are not counted. A topic that spikes but isn't answered well is your cue to add a Q&A pair or update a knowledge source.
Visitor satisfaction (CSAT)
Available on Advanced Pro.
The score comes from the end-of-chat survey, which you switch on or off per agent in Deploy, under Customer Satisfaction (CSAT). If the survey is off, the card says so and links straight to that switch, so an empty card never leaves you guessing whether nobody answered or nobody was asked.
On top of the metrics above, Advanced adds the satisfaction layer: the average CSAT score visitors gave, the share of replies that got a thumbs up, the running total of thumbs up and thumbs down, and the most common reasons behind a thumbs down. These count visitors' ratings only: your own ratings in Chat history never change them. Ratings given before visitor and team ratings were kept apart can't be attributed, so they are left out and the card says how many.
Use it to separate "my agent answered a lot of conversations" from "my agent answered them well". A high assumed resolution with a low CSAT usually means the agent is closing conversations without actually solving the visitor's problem.
Starter workspaces see conversation volume, response time, and top questions, but not the satisfaction breakdown. Assumed resolution shows as not measured on Starter, because the plan has no human handover to compare against.
On the Free plan the analytics page shows sample figures so you can see the shape of the report before you upgrade. Those are illustrative numbers, not your own. Your real conversation volume and response time appear once you are on Starter, assumed resolution once your plan includes human handover and it is switched on, and the satisfaction breakdown once you are on Advanced.
Export the report
Any insights view can be downloaded. Select Export, next to the time range picker, then choose a format. The export always covers the range currently on screen, so the file matches what you were looking at.
- Download CSV, the whole report as a spreadsheet: the three headline metrics with their trends, one row per day of conversation volume, your top questions with counts, and the satisfaction figures. It opens in Excel, Google Sheets and Numbers with no import step, and saves as
echo-insights-your-agent-2026-07-01-2026-07-14.csv. - Print / Save as PDF, opens a printable version of the report in a new tab and starts your browser's print dialog. Pick Save as PDF as the destination to keep a copy, or send it to a printer. If nothing opens, allow pop-ups for this site and try again.
Every date and time in an exported file is UTC, and the file records the moment it was generated. Days with no conversations are included as zero rows, so the daily series has no gaps when you chart it.
Satisfaction figures are part of the export on Advanced Pro for the preset 7, 14, 30, and 90 day windows. A custom start and end date exports everything else and states that CSAT covers trailing periods only, rather than quietly mixing two different windows.
Exports follow the same plan rules as the page itself. Starter and above download their own data, and Free workspaces are asked to upgrade instead of exporting the sample figures.
Still stuck? We'll help.
Every plan, including Free, comes with human support. If something isn't behaving the way you expect after following the steps above, send us a note and we'll get back to you promptly.