Timeline · Audit & Evidence
Timeline is the SwiftCase audit logger. Every action — AI or human — recorded with actor, timestamp, resource, outcome, and severity. Per case. Per agent. Per second. Exportable as a regulator-ready evidence pack.
The Data Model
Timeline records actions in a structured shape that handlers, managers, and regulators can all read.
actorThe named human or attributed AI agent that performed the action. Every entry has an owner — there is no anonymous activity in Timeline.
timestampPer-second precision, UTC. Both event-time and ingestion-time captured. Timezone-aware reporting available for multi-region operations.
resourceWhat was acted on — case ID, agent definition, workflow, field, document, conversation. Every action ties back to a structured object.
outcomeSuccess, failure, or partial. Includes the structured result data — what changed, what failed, what was returned. Not just “something happened.”
severityLow, medium, high, critical. Routine field updates are low; tool failures and escalation triggers are high; PII access and policy-guard breaches are critical.
filterableEvery dimension is queryable — by date range, actor, resource type, outcome, severity, or any combination. Fast on the case timeline; fast across the platform.
Example Timeline
An anonymised slice of a Timeline for one TL settlement. Every entry attributed. Every outcome captured. PII redacted for this example.
Set total-loss valuation on case CL-184729 (£8,450) with comparable market evidence pack attached.
Presented engineer-set value to customer on outbound voice call. Comparable evidence shared symmetrically (3 below, 4 above).
Customer agreed to value. Structured response captured: agreement_confirmed=true, hesitation_level=none, vulnerable_flag=none.
Drafted settlement-offer email. Status: pending_approval. Routed to handler queue for review.
Reviewed and approved settlement-offer email draft. Sent to customer at 14:37 UTC.
Escalation triggered: frustration detection on customer turn (call CL-184811). Routed to engineer Sarah J. with full conversation summary.
No anonymous AI actions. Every Switchboard entry references a specific agent definition (and version) and ties back to the named engineer or handler whose configuration it is executing. Personal accountability, mapped to actions, by design.
Why It Matters
FCA SMCR, Consumer Duty, FOS file requests, GDPR right-to-erasure — all of them ask for the same thing. Timeline answers it.
Senior Manager Conduct Rules expect personal accountability mapped to actions. Timeline provides exactly that mapping — the engineer who set a value, the handler who approved an email, the manager who signed off a threshold change. No “the AI did it.”
Consumer Duty asks for documentation that customers were treated fairly. Timeline produces that documentation as a side-effect of normal operation — what was offered, what was disclosed, what the customer said, who escalated when, and what the named human did about it.
When a case lands at the FOS, you hand over a structured evidence pack: full transcript, decisions with attributed humans, comparable evidence shown to the customer, escalation events, and the resolution path. Not a hunt through inboxes.
Every access to personal data is logged with actor, purpose, and outcome. Subject access requests can be served from Timeline. Right-to-erasure preserves audit hashes while removing PII. ICO inspection is a query, not a project.
Outputs
Same data, four audiences. Handlers see case timelines. Managers run cross-platform queries. Regulators get evidence packs. Engineers get webhooks.
Every action on a single case rendered as a chronological timeline. Used by handlers, managers, and complaint-investigators alike. No more “hunt the email thread.”
One-click export of a case’s full Timeline as a regulator-ready PDF. Filterable by date range or event type. Cryptographically signed with case hash for chain-of-custody.
Filter all activity across the tenant by actor, resource type, outcome, severity, or date. Find every escalation last quarter. Every failed tool call. Every PII access by a specific handler.
Stream Timeline events to your SIEM, data lake, or compliance tooling in real-time. Pre-built schemas for Splunk, Datadog, and generic webhook consumers.
FAQ
Default retention is seven years to align with FCA record-keeping expectations and FOS referral windows. Per-tenant retention policies can override this — longer for litigation hold, shorter for specific event types where regulatory requirements differ. Right-to-erasure under GDPR preserves audit hashes while removing PII content.
Yes. Timeline ships with structured event types for common actions (case create, field update, conversation, tool call, approval, escalation, etc.) and supports custom event types for tenant-specific workflows. Custom events get the same filtering, retention, and export treatment as built-in events.
Yes — explicitly. Every tool call records its outcome (success/failure/partial), every escalation records its trigger reason (low confidence, frustration, explicit request, error), and every provider-failover event during AI resilience routing is logged. Failures are not hidden; they are evidence.
Timeline events stream to webhooks in real-time with structured JSON payloads. Pre-built schemas for Splunk, Datadog, and generic SIEM ingestion are available. For higher-volume tenants we offer batched export to S3 / Azure Blob / GCS with deduplication and replay support.
Yes. Two paths: real-time webhook delivery for ingestion-as-it-happens, or a batched daily export to your blob storage with full schema metadata. Both preserve the full event structure including actor attribution and severity classification.
When a data subject erasure request is processed, Timeline preserves the audit record (action occurred, by whom, when, on what resource type) but removes PII content from the event payload. This satisfies GDPR Article 17 while preserving the regulatory audit trail. Cryptographic hashes ensure the audit log itself remains tamper-evident.
Application logs are for engineers debugging code. Timeline is structured business-action logging designed for regulators, handlers, and customers. Every entry has a named human or attributed AI actor, a structured resource reference, and a regulatory-relevant outcome. You can hand a Timeline export to the FOS; you cannot hand them a Splunk query.
Run a 30-day pilot. We’ll set up Timeline on the workflow you choose, and you’ll see the audit trail your existing tooling can’t produce. No platform migration. No long lock-in.