AI-powered email processing. Every incoming email classified, data extracted, and routed to the right place. Draft responses generated for approval. Your inbox, automated.
Email Intelligence
Not keyword matching. Not rule-based routing. AI that actually reads and understands your emails, then takes appropriate action.
AI reads every incoming email and classifies it by type. New enquiry, complaint, request for information, follow-up. Routed to the right place automatically.
Pull structured data from unstructured emails. Names, dates, reference numbers, amounts. Captured and stored in SwiftCase without manual entry.
Every classification comes with a confidence score. High confidence? Auto-process. Low confidence? Flag for human review. You set the thresholds.
AI selects the right response template based on email content. Personalised with extracted data. Ready to send or queue for approval.
AI-drafted responses always land in a handler approval queue: pending_approval → approved → sent (or rejected). Every outbound email traces back to a named human who clicked approve.
Emails linked to cases automatically. Create new cases from emails, or attach to existing ones. Full email history on every case.
Classification
AI understands intent, not just words. Same email types handled consistently, regardless of how customers phrase things.
First contact from potential customer
Customer needs help with something
Customer expressing dissatisfaction
Question about services or status
Attachments requiring processing
Reply to existing conversation
How It Works
Every email goes through the same intelligent pipeline. Classification, extraction, routing, and response. All logged and tracked in SwiftCase.
Automation
Emails routed to the right team based on content, sender, or subject. No manual triage needed.
Response time targets by email type. Track performance, alert before breaches. Never miss a deadline.
Filter spam before it reaches your AI. Flag suspicious content. Protect your team from malicious emails.
Draft responses queue for approval. Review, edit, approve, or reject. Batch processing for efficiency.
Approval Queue
Switchboard email never auto-sends. Every outbound email goes through a four-stage lifecycle, and every transition is attributed to a named human in Timeline.
receivedInbound email parsed, classified, linked to case.
pending_approvalSwitchboard composes a draft using your templates and case context. Routed to the handler queue.
approvedNamed handler reads the draft, edits if needed, clicks approve. Approval logged with handler ID and timestamp.
sentEmail sent to recipient under the approving handler’s attribution. Full audit trail in Timeline.
rejected — handler discards the draft and writes their own. Rejection reason captured.Every customer email proves a named handler reviewed and approved the content. Selective AI drafting with no human gate is a Consumer Duty risk; approval queues remove it.
When the FCA asks who approved a customer communication, the answer is in Timeline: a person, with a timestamp, on a specific case. Not “our AI did it.”
If a complaint reaches the FOS, the file shows the AI draft, the handler edit (if any), the approval timestamp, and the sent email. Every step attributed.
An AI that auto-sends emails can be tricked into writing things you would never approve. The queue catches every draft before it leaves the building.
Use Cases
FNOL emails processed automatically. Policyholder identified, claim details extracted, case created. Ready for handler review.
Support emails classified and routed. Simple queries auto-responded. Complex issues escalated with context.
New matter enquiries processed. Client details extracted, conflict check initiated, intake form pre-populated.
Invoice emails processed. Amounts extracted, PO numbers matched, approval workflows triggered automatically.
We'll show you how Switchboard processes real emails. Classification, extraction, routing, and response generation. Your inbox, transformed.