Skip to main content
This page is for the person answering a security review. It covers what WireDesk does with access, credentials, the chat widget and your data. For the technical detail, see the security model.

​
Access to your workspace

  • Your data stays in your workspace. Knowledge, conversations, contacts and leads are queried per workspace, behind a signed-in session. Every dashboard page checks that session on the server before it renders.
  • Three roles. Owner, admin and agent. Only the owner can invite people, change roles, see billing or close the account. Agents cannot change how the AI agent behaves. See team and roles.
  • Invitations are tied to an address. An invitation link only works for someone signed in with the address it was sent to. It expires after a week and can be revoked before it is used. Nothing is granted until the invitee presses Accept the invitation, so a link scanner opening the URL cannot accept it.
  • Changes come from the app itself. Requests that change something are refused unless they come from the WireDesk app.
  • Activity log. Changes to agents, channels, numbers, actions, knowledge, the widget, the spend cap, the plan and the team are recorded with who made them. Everyone in the workspace can read the log in Settings → Activity.

​
Signing in

  • Passwords are at least 8 characters. WireDesk stores a hash, never the password.
  • Changing your password signs you out on every other device.
  • Changing your sign-in address has to be confirmed from your current inbox, so a stolen session cannot move the account to a new address.
  • Password reset links work once and expire in an hour.
  • Sessions last seven days, and are extended while you keep using the app.
See sign in.

​
Credentials your actions use

API keys you save for API call actions are encrypted at rest with AES-256-GCM. After you save a key, it is never shown again and never given to the AI agent. It is only attached to the outgoing request.

​
The chat widget

A widget key is public by design: it sits in your page source. Three things stop it being misused:
  • Allowed domains. List up to 20 domains. The widget then refuses to load on any other site, and browsers refuse to show its chat frame inside any other page.
  • Verified visitors. For signed-in pages. Your server signs each user’s id with a secret only it and WireDesk hold, so a visitor cannot pretend to be someone else.
  • Rate limits. Each widget accepts up to 30 messages a minute from one visitor, and up to 30 a minute from one IP address. Past that, the visitor sees “You’re sending messages too quickly. Give it a moment.”
On email, the agent sends at most 8 replies an hour to one sender. This stops two auto-responders from answering each other forever. Later emails are still stored, but get no reply until the hour has passed. The spend cap is the final backstop. However much traffic arrives, what you pay past your plan cannot exceed a number you chose.

​
In transit and in the browser

  • All traffic is served over TLS. Browsers are told to use HTTPS only, for a year, on every subdomain.
  • The WireDesk app cannot be framed by other sites. The only part that can be framed is the widget’s chat page, and only on your allowed domains when you have set them.
  • Browsers are told not to guess content types, and to send only the origin as the referrer to other sites.

​
What we store

From the account owner and team: name and email address, a password hash (if you do not sign in with Google or GitHub), what you configure, and usage counts. Card details go to Stripe and never reach WireDesk. From your customers: whatever they say to your agent, and how to reach them back — a phone number on a call, a browser-scoped visitor id on your website, an email address in a thread. Plus anything the agent is configured to collect. Voice calls are transcribed. WireDesk stores the transcript, not the audio, so there is no recording to retrieve or lose.

​
Who else sees it

Only the processors the product needs in order to work — for generating replies, hosting and storage, phone numbers and SMS, email delivery, and payments — and only the part each one needs. The privacy page names them. WireDesk does not sell data, and does not use one customer’s conversations to improve another’s agent. Your knowledge base is scoped to your agents and never searched for anyone else. Region pinning is not offered today. If you need data kept in a specific jurisdiction, raise it before you sign up.

​
Retention, export and deletion

  • Conversations, contacts and collected data are kept while your account is open.
  • Diagnostic records of inbound calls are dropped after seven days.
  • You can export conversations, contacts, leads and collected data as CSV. See exports.
  • Closing your account releases your phone numbers, cancels your subscription, and deletes your agents, conversations, contacts and knowledge. See settings.

​
Cookies

The app sets cookies only to keep you signed in. There are no analytics or advertising cookies. The chat widget stores a visitor id in the browser on your site, so a visitor’s conversation survives a page reload. It identifies nobody by name.

​
Compliance

SOC 2 and formal DPAs are on the roadmap. They are not in place today. We will complete your security questionnaire and walk your team through the architecture. To start a review or report a vulnerability, use wiredesk.ai/security.