Ticket Form Mode

The Keva widget supports three modes — chat, form, and hybrid. The ticket form mode renders a structured submission form instead of (or alongside) the chat panel, with AI-powered intake baked in: paste an email thread or chat log and Keva extracts the subject, body, and customer info automatically.

When to use each mode

Install

The widget is installed the same way regardless of mode. Add the script tag to your site and set data-mode:

<script
  src="https://widget.keva.support/embed.js"
  data-api-key="keva_live_..."
  data-mode="form"
  data-primary-color="#3b82f6"
  async
></script>

Or programmatically

window.Keva.init({
  apiKey: 'keva_live_...',
  mode: 'hybrid',
  primaryColor: '#3b82f6',
  user: { email: 'customer@example.com', name: 'Jane Doe' },
  intakeModes: {
    paste: true,
    voice: true,
    screenshot: true,
    recording: false, // e.g. disable screen recording for privacy-sensitive deployments
  },
});

What the form looks like

1. Paste-anything block at the top — customers paste an email thread, chat log, or error message. A single "Extract ticket" button calls Keva's AI which fills in the subject, body, and customer info.
2. Voice dictation — a 🎤 Dictate button appears beside the Message label (Web Speech API, Chrome/Edge/Safari). Customers speak, their words stream into the textarea in real time.
3. Screenshots — customers can paste, upload, drag-drop, or capture screen screenshots inline. Up to 5 per ticket.
4. Screen recording — guided 3-minute screen recording with pre-flight privacy guidance, mic narration toggle, pause/resume, "+2 min" extension.
5. Structured fields — email, name, subject, message. The customer can review and edit what Keva extracted before submitting.
6. Submit — creates a ticket via POST /api/v1/widget/tickets with all attachments, and shows a success state.

Intake modes (per-feature toggles)

Each intake method can be independently turned on or off. All default to on; omit the intakeModes config entirely to enable everything.

window.Keva.init({
  apiKey: 'keva_live_...',
  mode: 'form',
  intakeModes: {
    paste: true,       // paste-and-extract (default)
    voice: true,       // voice dictation (default)
    screenshot: true,  // paste/upload/capture screenshots (default)
    recording: false,  // 3-min screen recording — disable for regulated industries
  },
});

When to disable specific modes:

• Screen recording off for medical, legal, or financial workflows where customers may inadvertently record sensitive screens.
• Screenshot off for similar reasons if recording is OK but images aren't.
• Voice off for sites where speech recognition is inappropriate (silent offices, libraries).
• Paste off for simpler workflows where you want customers to describe the issue themselves rather than pasting prior conversation.

Configuration

API key scopes

Your API key needs either tickets:write or widget:write scope to accept ticket submissions. Create a key in Settings → API Keys with the appropriate scope.

Events

Subscribe to widget events programmatically:

const keva = window.Keva.instance;
 
keva.on('form-extracted', (data) => {
  console.log('Keva filled these fields:', data.filled);
});
 
keva.on('form-submitted', (data) => {
  console.log('Ticket created:', data.ticketId);
});

Rate limits

To protect against abuse when embedded on public pages:

• 5 AI extractions per minute per API key
• 10 attachment uploads per hour per API key
• 10 ticket submissions per hour per IP address
• 20 ticket submissions per email per day

Customers hitting a limit see a friendly "please try again later" message.

Security & SOC 2

• All media uploads go directly to S3 via short-lived (10-min) presigned URLs
• Storage uses SSE-KMS encryption with a customer-managed key
• CloudTrail logs every upload + download for audit trail
• Retention: 2 years by default, matches your ticket retention policy
• Media flagged for PII (email, phone, card numbers) is stored in a separate, access-restricted path

See the Security section for full SOC 2 Type II details.

Troubleshooting