Tier model
The playbook is layered. User-entered facts win. AI-derived suggestions sit underneath. Market practice catches what is left. One mental model spans every page in the workspace.
The three tiers
| Tier | Surfaces as | Authority | Sources |
|---|---|---|---|
| Tier 1: User | User · Entered, User · Imported, AI · Verified | Highest | Admin manual entry, CSV import, and AI or market entries a human has verified. |
| Tier 2: AI | AI · Template, AI · Contract | Suggestion | Extracted from your templates, or derived from the corpus of verified obligations. |
| Tier 3: Market | Market Practice | Fallback | clmSpace’s curated market-practice corpus. Read-only. |
How a position rises
Every entry is labelled by where it came from. Verifying an entry in the Standards page marks it as verified and promotes it to its User form: an AI · Template entry becomes a user position verified from a template, a position derived from history becomes one verified from history, and a Market Practice entry becomes one verified from market. In each case the source of the position is preserved, so you always know what it grew out of.
Verified entries are Tier 1. They win specificity ties against any AI or market entry for the same scope.
What “winning” means
Whenever the platform needs the standard position for an obligation_type at a given scope (to compare an extracted obligation against, to inject into a drafting prompt, or to surface in the renewal pack) it resolves a single winner:
- Score every entry by specificity (count of matching scope dimensions).
- Higher score wins. Ties break by tier (User > AI > Market).
- Within a tier, more recent edits win.
The verify flow in practice
AI suggestions show up in the Standards page with an amber “Awaiting verification” treatment. Click Verify and they switch to the Tier 1 emerald treatment instantly. The same action is available to approved integrations through the Verify API.