AI Ask
An AI assistant that answers questions about board-level security using only Salvador Cloud's published content. Designed under our own AI Security Guardrails framework.
What it does
AI Ask answers questions about vCISO operating models, AI security guardrails, DORA readiness, cloud security architecture, and the regulatory frameworks Salvador Cloud advises on. It cites the Salvador Cloud article it drew the answer from, so you can verify the source.
Why it's not live yet
We don't ship AI features that don't pass our own guardrails. The feature is built (form, Worker, Vectorize index, KV rate-limit store) and operational; we are wrapping the pre-launch red-team in Q3 2026.
Security model (already documented)
- Input filter — adversarial inputs flagged before reaching the agent.
- Behaviour cage — agent has read-only access to the published-content corpus only. No tool calls; no PII access; no retention beyond the request lifetime.
- Output guard — every response checked against a known-bad-response library and a citation requirement before reaching the user.
- Rate limit — 20 requests/minute/IP via Cloudflare WAF.
- No PII storage — request bodies are not persisted; only aggregated event counts (Workers Analytics Engine).
Get notified at launch
Subscribe to the monthly briefing. AI Ask launch will be announced there.