What Can AI Read? — Machine Readability Must Follow Permission
How Forwork can design machine readability so AI understands Work Identity correctly while still respecting user-controlled permissions, visibility, and scope.
AI-first does not mean AI-everything
Structuring data for AI should improve understanding, not automatically expand access.
Step 1 — Separate structure from access
Data can be highly structured while access remains restricted.
Structured does not mean public.
Framework: Structure → Permission → Readability → Retrieval → Controlled Understanding
Structure — is the data organized clearly enough?
Permission — which entities may AI access?
Readability — what may be exposed in machine-readable form?
Retrieval — can AI retrieve the right permitted data?
Controlled Understanding — can AI understand correctly without exceeding scope?
Step 2 — Define machine-readable scope per entity
A public Profile may be readable. Private Evidence may not be. A selective Result may be readable only in an authorized session.
Step 3 — Avoid hidden access paths
Metadata, APIs, sitemaps, or indexing endpoints should not expose content the UI marks private.
Step 4 — Readability should reflect current permission
When a user changes visibility settings, the machine-readable layer should update accordingly.
Step 5 — Retrieval needs context
AI needs to know not only whether something is readable, but in which context: public web, signed-in user, shared link, or organization workspace.
Step 6 — AI summaries need provenance
If AI creates a summary, the system should know which sources, permissions, and data were used.
Applying this on Forwork
A useful model: Entity Structure → Permission Check → Machine Exposure → Retrieval Scope → Source-aware Answer. If permission fails, readability stops.
Conclusion
Do not only ask, “How do we make AI understand the Profile better?”
Ask: “How do we let AI understand the right permitted data, from the right sources, without exceeding user-selected visibility?”
Structure → Permission → Readability → Retrieval → Controlled Understanding.