Official Vault
Approved facts and policies
The Business Brain is a governed evidence system for founder judgment, official truth, temporal memory, customer context, policy, learning, and employee authority. Models can change. The company's approved meaning remains inspectable and yours.
Not one prompt. Not one vector index.
Raw conversations, uploaded files, websites, connectors, and employee observations begin as evidence, not truth. Provenance, authority, scope, temporal validity, conflict handling, and approval determine what can influence future work.
Approved facts and policies
Principles and tradeoffs
What was true, and when
Permissions and no-answer rules
What worked in reality
Context compilation
Choose a source to see how authority, exclusions, and role scope alter the live context pack.
The Brain assembles the smallest trusted packet this employee needs now.
Approved on 8 July by the workspace owner
Offer Saturday only inside the approved service radius.
What this changes for a small business
The Brain keeps hard-earned knowledge available, current, explainable, and reusable without forcing an SMB to build an internal AI team.
Founder judgment no longer has to be repeated separately to every employee.
New employees begin with current company truth, not scattered files and oral history.
Customers receive the same approved meaning across calls, chat, and email.
Useful corrections become governed company memory instead of disappearing in conversations.
Source priority
A customer message may reveal a useful pattern, but it cannot rewrite pricing or policy. Contradictions create an inspectable review item with both sources attached.
Cannot be displaced by a convenient answer
Approved rules, limits, and operating boundaries
Current documents, pricing, offers, and service facts
Approved principles, priorities, and judgment
Evaluated learning with evidence and version lineage
Customer and case facts for the active task
One parent Brain, role-specific projection
The parent Brain compiles a bounded child context for each employee. Company policy always wins; role knowledge and customer-private context remain scoped to the people and tasks allowed to use them.
Company truth, policy, source priority, approved memory
Offers, qualification, objections, pricing limits, follow-up, CRM tools
Scoped context packTroubleshooting, empathy, account facts, refund boundaries, handoff rules
Scoped context packAvailability, service rules, identity checks, booking, routing, reminders
Scoped context packTemporal Vault Graph
A normal knowledge base tends to flatten the latest text into one answer. Temporal memory keeps valid-from and valid-until windows, superseded facts, conflicts, source links, and the employees affected by a change.
Governed learning loop
A call, message, correction, tool result, or workflow outcome becomes immutable evidence.
The system extracts a possible fact, pattern, playbook, risk, or missing answer.
Source, scope, sensitivity, confidence, cost, and policy determine the safe disposition.
Behavior-changing proposals face tests before they can reach an active version.
High-risk or strategic change stays with the owner; low-risk automation needs explicit policy.
A new Brain version refreshes only the affected employee context packs.
Future outcomes test whether the new memory helped, failed, or should be superseded.
The exact previous approved state remains restorable.
Control is part of intelligence
Every meaningful fact can point back to its source and version.
Retrieval is tenant, role, department, customer, and task aware.
Contradictory evidence is quarantined for review, never averaged into certainty.
Behavior-changing proposals face tests before wider employee use.
Sensitive updates stay behind owner or policy review.
Version history restores the exact earlier approved state.