The knowledge desk that refuses to guess.
Load the documents your company already relies on. Ask in plain language. Every answer shows the exact passage it came from — and if your documents don’t prove it, the desk says so instead of guessing.
Watch a real question
Every answer is weighed before you see it.
A second, independent AI checks every part of the draft against the documents. Only a fully proven answer is shown.
a scripted example over the fictional Meridian corpus example
The part no other tool does
When the documents don’t prove it, it says so.
No confident guesses. The desk declines, shows the closest passages it found, and leaves the decision with you.
a scripted example — the honest refusal, at full ceremony example
The scoreboard
Checked on real graded questions, in the open.
How often the desk answers when it should, how often it correctly declines — measured on stored runs anyone can inspect.
0%
answered and proven
from stored run #6
0%
correctly declined
from stored run #6
0%
declined when it could have answered
from stored run #6
live numbers from the newest stored evaluation run — open the scoreboard
What the desk knows
Your whole back-file, on one living map.
Every document a point, every connection a line — versions, references, and the documents that contradict each other, found automatically.
an example map — the live corpus map arrives with the knowledge layer example
Built from blocks
Assemble the desk your business needs.
The running pieces are marked BUILT — watch them work. The rest we build on request, tailored to your documents, your rules, and the systems you already use.
The trust core
The honesty check
A second, independent AI checks every part of an answer against your documents before you see it.
Checking strictness tuned to your risk — a clinic is not a sales team.
watch it workProof on every answer
Every answer opens to the exact place in the exact document it came from.
Your documents, your numbering, your language.
watch it workThe honest “I don’t know”
Unproven answers are withheld — you get the closest passages instead of a guess.
The refusal message and the escalation path are written for your team.
watch it workThe scoreboard
Public, reproducible numbers on real graded questions — answered, proven, declined.
A private scoreboard on your own questions comes with every engagement.
watch it workLoading knowledge
Drop in your documents
Drop files, say in your own words what they are, confirm the plan — then ask.
The setup proposals learn your document types.
PDF, office files, web pages
Contracts and manuals in the formats they already live in.
Your formats first; scanned paper on request.
Connected sources
The desk reads straight from SharePoint, Drive, Notion, or Confluence and stays current.
We wire your two or three systems — not a 275-connector zoo.
Scheduled re-reading
Documents re-read on a schedule, so a changed policy changes the answers.
Protocols weekly, contracts monthly — your cadence.
Where your team asks
The web desk
Ask in plain language; get the proven answer or the honest no.
Your wording, your example questions.
watch it workAsk from Slack or Teams
Questions answered where the team already talks, proof attached.
Your workspace, your channels, your escalation rules.
Department desks
An HR desk, a contracts desk, an on-call desk — each seeing only its documents.
Scope, wording, and refusal path per team.
Share an answer
Any proven answer becomes a link a colleague can open — proof included.
Access follows your team boundaries.
Staying current
The knowledge map
A living map of everything the desk knows, growing as documents load.
Colored by your document families.
Corpus health
Flags what nobody cites, documents that contradict each other, and superseded versions.
Health rules that match your document lifecycle.
Verified by your expert
Your people stamp documents “verified” with an expiry; the desk prefers them and nags on expiry.
Your verifiers, your expiry policy.
Whose desk it is
Invites and the test budget
Access by invite, spending a real owner-funded budget with a hard cap in code.
Evaluation budgets sized per prospect.
Per-team access
Different groups see answers from different document sets — nothing leaks.
Mapped to your real teams.
Runs on your own machine
The whole desk — including the AI — on a machine in your building. Nothing leaves.
We size and set up the hardware.
Your choice of AI
You choose the models. The price follows.
Run on capable free-pool AI, on paid frontier AI, or entirely on hardware you own — the checking never gets weaker, whichever you choose.
Budget
measured$0.00 per answer
the public demo runs this tier; the checker is still a different AI than the writer
Quality
estimateabout $0.01–0.05 per answer
paid frontier AI from two different vendors — the checker never grades its own homework
Private
on requestruns on hardware you own
the whole desk on a machine in your building — your documents never leave
Ask about thisThe honest comparison
Why not just use the big tools?
Use them if they fit. Here is the one thing they don’t do.
| Tool | What it does well | What it won’t do |
|---|---|---|
| Glean | Company-wide search and assistant for large enterprises, with strong citations. | Its own docs say ungrounded answers still reach users — it monitors quality afterward; it never refuses. Sales-led, ~100-seat economics. |
| Guru | A verified company wiki — humans stamp documents as trusted. | Verification is a badge on documents, not a check on answers; its guardrails are off by default and fail open when the checker is down. |
| Onyx | Open-source chat over your company tools; you can self-host it. | Bets everything on good retrieval — no check that the answer is actually proven, and the permission layer is paywalled. |
| Do it yourself | Free open tools can chat with your documents on your hardware. | A raw tool, not a service: no checking layer, no scoreboard, and someone in-house has to build, tune, and maintain it. |
| Proofdesk | Answers proven against your documents, or honestly declined — with a public scoreboard. | It will not show you a guess. That is the point. |
The next step
See it run on your own documents.
An invite comes with a real, owner-funded test budget. You watch the desk answer — and refuse — on your material before you commit to anything.
What the manual way costs
For a 20-person firm, digging answers out of contracts and policies by hand runs to roughly $14,000 a year of interrupted time — before counting the one missed clause that auto-renews a bad rate. A worked example, not a measurement: your numbers replace these.
worked example
Your documents, and your team’s corrections, make it smarter over time — and it refuses to guess. It becomes your system, not a rented tool.