Company Brain Engine
The knowledge engine underneath: captures knowledge from conversations and permitted activity, searches context directly, and synthesises review-style summaries on demand.
Company Expertise
From data to memory.
The governed body of what the company knows: mined expertise, knowledge captured from real work, source-backed facts, context briefs, and a living wiki, the live operating memory layer agents and people draw on.
Agents fail because context is fragmented, not because models are weak.
Critical context lives across CRM, ERP, ticketing, documents, meetings, emails, and people's heads. It is scattered, partially stale, and unscoped, and agents either cannot find it or use it where they should not.
Company Expertise is the bridge between what the organisation knows and what governed agent teams can safely use. It is not a document index, not generic RAG, and not a chat history archive. Company Brain is the engine underneath: it captures, organises, searches, and synthesises Company Expertise under policy.
Inside Company Expertise
How Company Expertise works
Six layers turn fragmented context into a living, governed memory.
Organise expertise context
Stack expertise packages with related company facts, approval paths, and operating rules, each scoped to where it actually applies.
Separate scope levels
Keep company-wide facts distinct from expertise-specific context and personal preferences, by design, not by convention.
Connect to live systems
Link memory to CRM, ERP, ticketing, and internal tools so context stays current without manual refresh cycles.
Ground every fact in source
Anchor context in evidence (meetings, documents, corrections, approvals) so agents can cite their reasoning when asked.
Enforce access boundaries
Define which agents, channels, and people can read, learn from, or update each layer of memory.
Surface for governance
Expose Trust Review, the Review hub, and Brain Metrics so admins always understand what agents are seeing, how fresh it is, and why.
What Company Expertise contains
Scoped, source-backed, and inspectable, not a flat document dump.
- Mined expertise with interview evidence
- Reference facts, activity evidence & derived insights
- Company memory (org-wide facts)
- Personal memory & continuity
- Context briefs & concept network
- The living wiki: projected & authored pages
- Trust states & review queues
- Agent & channel permission boundaries
- Evidence trails for audit
- Tenant & personal document libraries
Capabilities
Six engines keep memory scoped, current, and governable.
Memory Scope Engine
Organises context into company, expertise, and personal layers with automatic isolation rules. Personal memory stays personal, with per-item forget and no admin in the loop.
The Living Wiki
The authoritative, readable layer: pages synthesised from captured knowledge that humans accept, plus authored pages people write, with corrections, contradiction detection, and structural curation.
Trust Model
Every governed artifact carries a trust state: verified, review due, trust at risk, or quarantined. The platform only ever lowers trust on its own; only a human restores it.
Channel Read & Write Policy
Knowledge access is set where conversations happen: each channel controls what agents may read and whether conversations may update memory, enforced server-side and failing closed.
Company GPT
The member Workspace chat experience over permitted Company Expertise: one interface with personal memory, documents, connected tools, and the same governance as every other surface.
Why this is different
Scoped, not sprawling
Context is organised by expertise, company level, and personal scope, not a flat document dump that bleeds across boundaries.
Action-ready, not just retrievable
Designed for agents to use in decisions and drafts, not just to surface answers in a search.
Evidence-visible, not hidden
Users and admins can review source, freshness, and provenance when needed, without noisy detail by default.
Governed by policy, not convention
The same Company Expertise supports different read and learning rules for internal teams, support agents, embedded widgets, and backend automations.
See Verti run on your operations.
Verti connects expert judgment, live context, governed action, and learning loops into one production system.