Skip to main content

​
Pricing and usage

No. The Free plan needs no card and includes 500 replies or 50 voice minutes a month, with real customers as well as in testing. You only need a card to keep answering past that, or to move to a paid plan.
Two things: how much usage is included each month, and how much help you get. Every feature is on every plan, including Free: chat, voice and email, unlimited agents, custom API actions, roles, the activity log, analytics and the spend cap.See Plans and limits.
0.02peransweredchatmessageoremail,and0.02 per answered chat message or email, and 0.20 per started voice minute. Replies and minutes draw on the same monthly allowance, so one voice minute uses as much as ten replies. Past the allowance, usage is billed at the same rates, up to your spend cap.
No. The allowance resets on the first of each month, on monthly and yearly billing alike.
The agent stops answering billable conversations until the month rolls over, unless you add a card. With a card on file, it keeps going at list rates, up to the spend cap you set.
Yes. Every workspace has a hard monthly cap on what you pay beyond your plan. It is $50 by default, and you can change it. When the cap is reached, the agent stops answering billable conversations until you raise it. See Spend cap.
No. There are no seats on any plan. Ten people working the inbox cost the same as one.
Yes. That answer is a reply like any other, and it counts against your allowance. It is also a signal: an agent that says it a lot has a knowledge gap. See The agent says it does not know.

​
The chat widget

Yes. The key identifies the widget. It lets someone chat with that one agent and does not give access to anything else. Three things stop abuse: allowed domains, rate limits on the chat endpoints, and your spend cap. See Security model.
widget.js is about 6 KB gzipped and loads asynchronously. It does not build the chat iframe until a visitor hovers over, focuses, touches or clicks the launcher. On a page where nobody opens the chat, that one small script is all it costs.
Yes. Whether the panel is open is kept for the browser tab, so clicking through to another page does not close a conversation in progress. A visitor who comes back within 24 hours finds the same conversation. After that, they start fresh. They can also choose View recent chats in the panel’s menu to reopen an earlier conversation, or Start a new chat.
Yes, with no JavaScript. Add data-wiredesk-open to any element, or give it a value to type a question in for the visitor: data-wiredesk-open="I need help with billing". From code, call wiredesk("open"). See JavaScript API.
The browser is refusing to show the conversation frame. Either your site’s Content-Security-Policy does not allow frames from https://wiredesk.ai, or the page’s domain is not on the widget’s allowed list. The console names which one. See The chat bubble is not showing up.
The allowed-domains list almost certainly has one host and not the other. A bare domain covers its subdomains, but it does not cover a different domain.

​
Integrations

Not today. WireDesk has no public REST API and no outbound webhooks. To send data to your systems during a conversation, use an API call action: the agent calls an HTTPS endpoint of yours. To get history out afterwards, use the CSV exports for conversations, contacts, leads and collected data.
Two ways. Export them as CSV, which needs nothing built. Or add an API call action that sends them to your own endpoint, if you want them to arrive live.
Yes. An API call action lets the agent call an HTTPS endpoint of yours, such as an order lookup or a stock check, and answer from the response. See API calls.

​
Phone and handover

No. You keep it and forward it to the agent’s number. Your published number stays with your current carrier.
No. Calls are routed by the number dialled, so each number belongs to one agent. If you want different answers on different lines, give each line its own number.
Yes. Test by voice, at the top of every agent page, starts a voice call in your browser. It does not need a phone number.
The request times out, and the agent tells the customer that nobody was free rather than leaving them waiting. The conversation stays in your history as an unanswered request. See Handover.
Yes. Your team replies as themselves, and a person’s replies are marked as theirs in the chat.

​
Knowledge and answers

It is told to answer from what it finds in your knowledge and to say when it does not know. If you see a confident answer with no source, it almost always comes from a fact typed into the agent’s instructions. See Fix a wrong answer.
PDF, Word, Excel, CSV, HTML or plain text, under Knowledge → Upload a file. See Files.
Usually one of three reasons: the source is still processing, the page could not be reached without a login, or the sentence is buried in a page that is mostly header and footer. A short note that states the fact plainly fixes all three.

​
Team and security

There are three roles:
  • Owner: everything, including billing and deleting the workspace.
  • Admin: agents, knowledge, numbers and the inbox. Not billing.
  • Agent: the inbox and conversations. Cannot change how the agent works.
Only the owner can invite, remove or change the role of teammates. See Team and roles.
Not yet. SOC 2 and formal DPAs are on the roadmap. In the meantime, we will complete your security questionnaire and walk your team through the architecture. Start at wiredesk.ai/security, and see Security model for how the widget is secured.