Skip to content

[Feature] Identity layer — infer goals, preferences, and decision tendencies #21

@rahilp

Description

@rahilp

Why
A memory system that just stores and retrieves is a database. One that understands who you are from what you've stored — your recurring goals, how you make decisions, what you care about — is something more useful. This item is about building that inference layer on top of the raw memory store.

What

  • Goal inference: from stored memories, identify recurring themes that represent ongoing goals
  • Preference detection: extract stated and implied preferences ("I don't want X", "I prefer Y")
  • Decision pattern recognition: identify how decisions tend to get made (fast vs deliberate, solo vs collaborative)
  • Communication style: infer preferred level of detail, tone, and format from stored content
    Implementation notes
  • Runs as a periodic background job, not at capture time
  • Output stored as a special identity tag entries — not mixed into regular recall unless explicitly requested
  • Needs careful UX: surface inferences as "here's what I think I know about you" with easy correction
  • Do this last — it's high risk of feeling invasive if the rest of the system isn't solid first

Metadata

Metadata

Assignees

No one assigned

    Labels

    Projects

    Status
    Todo

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions