Reports
How well your brand's AI agent answers, and a browsable list of every answer it has given.
Where this is
Brand → (your brand) → AI agent → Reports. One screen holds the whole answer-quality picture: three summary figures, the conversations Yaplet carried on its own, and — at the bottom — Answers, the browsable list of every individual reply.
Everything here is about this brand's AI agent, over the date range you choose in the top right. That one range drives the figures, the conversation list and the Answers list, and it starts on the last month. Opening the screen needs the AI agent permission.
The three figures
Each figure is a tile with its share in large type and the count underneath it, such as "12 of 40 questions". Point at a tile to read what it counts.
Successful
Of the real questions asked, the share Yaplet handled — answered from your content, connected a requested person, or ran a workflow. Small talk is not counted.
Failed
Real questions Yaplet could not answer — missing knowledge, a hand-off it chose itself, or an error.
Small talk
Messages that asked nothing — greetings, thanks, acknowledgments — and hand-off checks. Shown as a share of all messages, and left out of the other two rates.
Conversations handled by Yaplet
Under the figures is the list of conversations Yaplet carried without a person stepping in, with the columns Chat ID, Handled by Yaplet, Answers and Last activity. Its title shows what share of all conversations in the range those were, for example "Conversations handled by Yaplet: 72%".
Conversations that needed no one from your team.
One long-running chat is split into segments when its messages are more than 3 hours apart, and each segment is counted separately.
The badges on each row break that conversation down into successful, small-talk and failed replies.
Selecting a chat ID opens that conversation in the inbox, in a new tab.
Answers
The Answers section at the bottom lists every reply this agent has written, newest first, in the columns Question, Answer, Date, Outcome and Response time. It has a Question search and an Answer search, an Outcome filter, a Partial answers filter (replies that covered only some of the topics a message asked about) and the date range from the top of the page. Each row carries an outcome badge; point at it to read what it means:
The AI answered the question using content retrieved from the brand's knowledge, or an API tool. (Counts as successful)
The AI handled a message that needed no content — a greeting or a simple pleasantry. (Counts as small talk)
The AI did not have enough information to answer the question. (Counts as failed)
The visitor explicitly asked for a person, and the AI handed over. (Counts as successful — the AI did the right thing)
The AI found nothing relevant, or a tool failed, so it handed over by itself. This one points at a gap in your knowledge. (Counts as failed)
The AI started a workflow that resolved the turn. (Counts as successful)
The visitor asked for a person, and before connecting them Yaplet checked whether it could solve the issue itself — see Let Yaplet try first on AI agent settings. The row reads Hand-off check with a one-line summary, and its badge says how the check ended: Yaplet could help, Handed to a person or Asked what they need. (Counts as small talk — never as successful or failed)
Something went wrong while the answer was being generated. (Counts as failed)
The Outcome filter names the groups a little differently: Answered, Neutral, No answer, Agent handoff, Workflow and Error. No answer is the Failed badge, Agent handoff covers both Person requested and Handoff (no context), and Neutral covers small talk and hand-off checks.
The Response time column shows how long the visitor waited — from their message coming in to the answer being ready. Click the header to sort by it and find your slowest answers. Times under 15 seconds are shown normally, from 15 seconds they turn orange (the visitor notices the wait), and from 30 seconds red (slow). Answers from before 20 October 2026 show —, because response times are recorded from that date.
Inspecting a single answer
Use the Details action on a row to expand it. Open conversation, at the top of the expanded row, opens the chat in the inbox in a new tab.
Conversation
The visitor's question and the AI's reply, rendered exactly as the visitor saw them — including any product carousel the AI attached. When a workflow's Send to the AI agent step started the turn, the question side is labeled Workflow hand-back instead of Visitor: the text there is the step's instruction, which the visitor never typed.
Details
Six fields summarize the answer at a glance:
| Field | Meaning |
|---|---|
| Outcome | The badge described above. |
| Channel | Where it happened — web chat, Facebook, Instagram, email or phone. |
| Grounded | Whether the reply was built on content found in the brand's knowledge. |
| Model | The AI model that wrote it. |
| Tokens | Total tokens the turn consumed. |
| Response time | How long the visitor waited for the answer, colored the same way as in the list. |
Where the time went. Under the six fields, a bar shows how the response time was spent, in the order it happens:
- Preparing — getting the conversation ready before the AI starts.
- Searching — the time spent looking things up in the brand's knowledge and products.
- AI thinking and writing — everything else: the AI deciding what to do and writing the answer, plus any API tool it called.
Hover a part of the bar to see its exact time. For web chats, where the answer appears word by word, a line also says when the visitor saw the first words — for example "First words after 4.2 s".
Below the fields and the time bar, one or two lines explain how the reply came about when the steps alone would not tell you:
- why a turn has no search step — the message was a plain greeting handled without a search; or the search found nothing and the AI judged the reply needed no company facts (small talk, a follow-up); or it found nothing and said so;
- for a message that asked about several topics — how many were found, and which ones the knowledge base had nothing on.
Steps
A numbered timeline of what the AI did before replying. Each step shows a status and opens for detail:
- Knowledge search — the search it ran and the pieces of content it matched, each with its source title and a relevance score. Open one to read the text itself. When the AI searched in a different language from the one the visitor wrote in, the step says so. A message with several topics shows one search per topic, labeled part 1 of 2, part 2 of 2; at the Deep search depth the extra phrasings appear as alternative phrasing; and when the AI's own query missed and it retried with the visitor's exact words, the step says retried with the visitor's own words. Each search also shows how long it took, and a This search: row breaks that time into its stages — Translation, Setup, Embedding, Database, Ranking, Claim check and Products & rules (stages that took almost no time are left out). When one stage took most of a slow search, it is highlighted.
- API tool — an external tool the AI called, with the data sent and received.
- Workflow — a workflow the AI started.
- AI-written message — a workflow message step with Let the AI say it on. Under Raw data, the input shows what you typed and the output what the AI said. A failed one means your typed text went out unchanged, and the error says why — for example, the visitor had used their daily answers, or the AI was paused for them.
What these numbers leave out
Improving what you see here
Read the failures first
Filter Answers by the No answer and Agent handoff outcomes. The rows badged Failed and Handoff (no context) are the questions your content does not cover yet.
Open the steps
Expanding an answer shows what the AI searched for and what it found — the fastest way to tell "nothing matched" apart from "the wrong article matched".
Fill the gap
Add what is missing on Brand → Knowledge, or on Brand → Products when the question was about your catalog. Anything you add becomes searchable straight away.
Decide what happens when it cannot answer
When Yaplet has no answer lives on AI agent settings, in the Human hand-off section. Set it to Connect to a person if you would rather unanswered questions reach your team than be politely declined: whenever someone from your team is online, the visitor is connected to them at once, and when nobody is, the visitor is told so and can leave a message for the team.