It can answer. It can summarize. It can sound impressively confident about your cloud bill. What it can't do is operate inside your business — your policies, your terminology, your guardrails.
That's the gap most "enterprise AI" launches in 2026 are quietly hoping you won't notice.
We just closed it for Billy.
Billy is Finout's AI FinOps assistant. Read-only, opted-in by an admin, audit-logged on every enable/disable, scoped to whatever RBAC and ACLs your user already has, running on Claude Sonnet inside AWS Bedrock — data never leaves the perimeter, never trains a model. That's the floor. That's what enterprise-grade is supposed to mean before the word stops meaning anything.
Today we're adding the layer that turns Billy from a clever chatbot into a member of your FinOps team: AI Account Guidelines.
Admins write the rules. Billy treats them as ground truth.
"Below 20% headroom on production RDS is never acceptable, regardless of utilization." "When engineers say 'infra' they mean the Platform Engineering team." "Never surface recommendations that touch backup=critical without DBA acknowledgment."
Business context. FinOps policy. Organizational glossary. Compliance constraints. Your language, your boundaries, applied to every answer Billy gives — to every user, every time.
Here's the part vendors don't want to say out loud: an AI assistant without this layer isn't enterprise-ready. It's a demo. A nice one, sometimes. But still a demo.
The question isn't whether your FinOps assistant talks. It's whether it knows the rules of your house.
Billy now does.