Workflow Drilldown: 7. Archive
Archive is what makes AI defensible. If a firm cannot reconstruct what was generated, reviewed, approved, sent, and relied upon, it cannot supervise AI safely.
What may need archiving
- Client-facing AI-generated emails, reports, summaries, and chatbot transcripts.
- Meeting transcripts and summaries used in planning or advice.
- Marketing drafts and approvals.
- Prompts and outputs where they support a recommendation or regulated communication.
- AI tool approvals, testing evidence, vendor due diligence, and policy exceptions.
- Human review/approval evidence.
Why archiving matters
AI creates many intermediate artifacts. Some are disposable drafts; others become books and records, compliance evidence, or source material for advice. Firms need policies defining which artifacts are retained, for how long, and in which system.
Good archive design
Final client notes and facts belong in CRM/planning systems, not scattered in AI chat history.
Audit trailRecord who reviewed, edited, approved, and sent AI-assisted content.
SearchabilityArchived content should be discoverable by client, date, advisor, topic, and communication type.
Retention/deletionRetention should meet regulatory and firm policy requirements without keeping unnecessary sensitive data forever.
Common mistake
Firms adopt AI tools that save time but cannot export records, integrate with archives, show approval history, or enforce retention. That creates a supervision problem later.