Inventory and ingestion
Every document source found, classified, brought in.
What's included
Connectors to the places documents live; classification by type, owner and business area; duplicates and drafts identified.
One place for what the company knows.
Documents, policies and decisions — permissioned, versioned, cited.
- The answer exists in a document nobody can find.
- Three versions of the policy; nobody knows which is current.
- The AI answered from an outdated PDF.
10B+messages · founder's prior work
6warehouse domains · founder's prior work
239tech stacks audited · our tooling
Five modules
Structure and permissions
One taxonomy; access that mirrors how the company is organised.
What's included
A taxonomy agreed with you; permissions carried over from the source systems and enforced on every read, by people and by agents.
Versioning and currency
Dates and owners on every item; stale content flagged.
What's included
Effective dates and supersession; owners and review reminders; anything past its review date marked stale and excluded from answers.
Retrieval for people and agents
Search for people, cited answers for agents.
What's included
Search with permissions applied; an interface agents use for FAQs, policies and procedures, returning the passage and its version. Facts — orders, balances, inventory — are never answered from here; they come from the warehouse.
Maintenance
New sources and stale-content reports. Monthly.
What's included
A monthly report of what was added, what went stale and what nobody owns; new connectors as sources appear.
How it works
- CollectEvery document source
- ClassifyType, owner, area
- PermissionMirrors the org
- VersionDates, owners, review cycles
- ServePeople and agents, with citations
- ReviewStale flagged monthly
Facts live next door, in the Data Foundation · knowledge answers "what is the policy", the warehouse answers "what is the number"
What we connect
- Google Drive
- SharePoint & OneDrive
- Notion
- Confluence
- Slack & Teams
- Ticketing
- PDFs & scans
Which sources come first is decided in the diagnostic.
Your data stays yours
- Your environment
- Read-only by default
- Your model channel
- In the contract
Permissioned. Versioned. Cited.
How this compares with wikis, enterprise search and chatbots over documents
| Wiki tools | Enterprise search | Chatbots over documents | Unvessel | |
|---|---|---|---|---|
| Covers | What people write there | What it can index | What was uploaded | Every source, classified |
| Currency | Manual | None | None | Dates, owners, stale flagged |
| Permissions | Its own | Usually | Often flattened | Mirrors the source systems |
| Answers | Pages | Links | Sometimes invented | Cited passage and version; facts from the warehouse |
| Where it lives | The vendor | The vendor | The vendor | Your environment |
How to start
- DiagnosticIncluded in the Data Health Diagnostic
- First sourcesFixed price
- MaintenanceMonthly
- We don't put facts in the knowledge base; numbers come from the warehouse.
- We don't serve answers without a source and a version.
- We don't flatten permissions to make search easier.
Questions
Is this a chatbot?
No. It is the layer a chatbot or an agent reads from — with permissions, versions and citations. Which interface sits on top is your choice.
Do we need the Data Foundation first?
Not for the knowledge base itself. You need it the moment an agent has to answer with numbers as well as policies.
Where does it live?
Your environment. Documents are not copied to us.
How is it priced?
Diagnostic at a fixed fee; sources at a fixed price; maintenance monthly.
Past work is on the case studies page.
Start with a diagnostic
Tell us where your documents live and who needs to find them. We'll reply with the scope and the fee — or tell you honestly if we're not the right firm.
info@unvessel.ai