Policies
Define what the AI may and may not say about your company — grounded in a corpus, cited, with an approved/forbidden claim list. The Gate enforces it. · 4 min read
Policies is where you set the boundary of what the AI may say. The model the product is built on: a corpus is the source of truth — the AI may state anything grounded in it, must cite it, and may say nothing the corpus can’t back.
The corpus is the boundary
Add the documents you stand behind (one-pager, case studies, security, pricing…). Each shipped line is cited back to one of them; anything ungrounded is blocked. The seed docs carry real (synthetic) content — click View to read exactly what may and may not be cited from each.
The six sections
| Corpus | The source documents the AI may cite. Not in the corpus → not sayable. |
| Claims | ✅ approved claims (each with a citation) and ⛔ a cannot-say deny-list. |
| Provability | The Gate — grounded-in-corpus + cited, deny-list honored, no fabrication. Inviolable. |
| Disclosure | What a 1:1 may DO with a fact by how you learned it — say / allude / hold. |
| Anti-slop | Kills the “a bot wrote this” fingerprint — AI-tells, filler, blast-y structure. |
| Data & consent | Sensitive topics never referenced, and whether consent is required. |
Tenant-editable vs platform-inviolable
Corpus, Claims, Disclosure, Anti-slop additions, and Data are tenant-editable — your edits persist per tenant. Provability is the Gate’s vetoes: a tenant can add stricter rules but can never loosen “can’t say what it can’t prove.” Note: tone / brand-voice is a style choice that gates nothing, so it is deliberately not a policy.