This is the persona the product is designed around, not one it was retrofitted for.
You built it for them. Let them watch it work.
If you ship conversational AI for clients, your problem is not observability. It is that the person paying you has no window into the thing they bought, and no structured way to tell you when it misbehaves.
If more than one of these is familiar, the rest of this page is for you.
The month you keep repeating.
The report
You screenshot metrics into a slide deck, again. It takes an evening and it is out of date by the time you send it.
The silence
Nobody mentions the agent. You have no idea whether it is going well or whether they have stopped using it.
The message
“The bot has felt off lately.” No date, no conversation, nothing you can reproduce.
The renewal
You are asked to justify the retainer with numbers you assembled by hand.
One account, one dashboard, every client’s brand.
Each agent gets its own subdomain, logos, colours and visibility rules. Your client opens something with their name on it, and never sees another client’s data or ours. Agents are not rationed by plan, run one per client, five per client, or a shared one across a whole account, the branding and the visibility rules are set per agent either way.
Edit any swatch, both themes are set independently.
Slot 1 tracks the primary colour.
Slot 1 tracks the primary colour.
Because it carries no AgentSight branding anywhere your client can see, the whole dashboard, the reports and the ticket board can be presented as your own in a proposal or a sales call, without a single caveat about whose infrastructure is running underneath.
How white-label worksThis is the part that changes the relationship, not just the reporting.
“The bot feels off” becomes a ticket.
Your client reads the transcript, finds the exchange that went wrong, and raises it against that exchange, with a priority, an environment and a thread where you both work it out. No more reconstructing a complaint from a Slack message.
Tickets are on every plan, free included. They are the reason a client keeps opening the dashboard instead of forgetting it exists.
See the loopAgent loses context after escalation handoff|
“The bot has felt off lately.”
Client flagged it happened twice more this week, always right after a transfer.
Found it, the summarizer was firing before transfer context got appended. Fix is up in staging.
They confirmed the next three sessions looked clean.
|
The retainer justifies itself.
Save a dashboard snapshot with its date range intact and export a branded PDF from a per-agent template. The renewal conversation stops being a scramble and starts being a document you already have.
Reports & PDF exportQuestions from agencies.
Can we manage several clients from one account?
Yes. Every agent has its own branding, environments, visibility flags, tokens and tickets. One login, separate worlds.
Will our client see that it is AgentSight?
On a branded subdomain they see your app name, your logos and your colours. The embed goes further, it lives inside your own product.
What if a client should not see personal data?
Eight per-conversation fields can be hidden from the client-facing view, and in the embed personal data is off by default.
How long does it take to wire up?
An afternoon. init(), a conversation, a turn, and the two messages, model calls inside the turn are captured with no extra code.
Can we bring history from before we switched?
Yes, import from a JSON export, so your client's dashboard does not start at zero on day one.
Stop screenshotting metrics into slide decks.
- Full dashboard on the free plan
- No card required
- Model spans never carry prompts
Pulled the trace for conv_8f2a1c, context resets right after the handoff tool call. Digging in.