About Services Tools Blog Contact us

Staff Knowledge Inbox

Staff paste a note or drop a file, and it is filed automatically into the same private search as the rest of the company record.

  • Knowledge Base & SOPs
  • Professional services
  • Pipeline
  • Product
  • Local/private
  • 2026

The problem

Decisions and fixes never reached the company record, so search knew less than the team did. Filing a note meant stopping work to do it, and most people did not. Sensitive text was as likely to be saved as a useful fix.

What it does

A signed-in form takes pasted text or a file, with an optional hint, and accepts it at once. Qwen, run through Ollama, drops duplicates and sensitive content, then works out a title, a type, tags and an access list no wider than the sender's groups. Accepted text is split up and prepared for search like any other source. The inbox is specified alongside the company record, and this page does not add anything to it.

Outcomes

  • A note is filed without a separate knowledge task.
  • Access stays within the sender's groups.
  • Duplicates and sensitive drops are refused before they are saved.
  • The contributor sees a sample question as confirmation.

Simplified mockup

Interactive preview with sample data. Nothing is sent to a server.

File a note

Ready

Signed in as Amina Okonkwo, consulting desk. A form token is already on this page. Access cannot be wider than her groups.

Note
Drop pricing-exception.txt, or click to attach it
Ingestion
  1. Queued
  2. Triaged
  3. Titled
  4. Embedded
Attach the file, then file the note.
Known duplicate do not index

Please index this again: hold the August day rate for NQ-4418.

Queue

Waiting
0
Ingested today
0
Refused
0
WhenTitleKindAccessStatus
09:05 Bank holiday opening note facilities ingested
08:22 Bonus notes note not stored sensitive
Yesterday Day rate to invoice procedure consulting, finance ingested

Sample retrieval

File a note to see the question it can answer. Nothing from this drop is in the index yet.

How it works

  1. Open the staff drop page behind a sign-in and a security code.
  2. Paste a note or drop a file, with an optional one-line hint.
  3. Accept the submission at once and show it as pending.
  4. Check for duplicates and sensitive content, then work out the title, tags and access.
  5. Prepare accepted text for search in the shared company record.
  6. Show the contributor what was stored and one sample question it can now answer.

Pipeline

flowchart LR
  Drop["Paste or file"] --> Queue["Queue"]
  Queue --> Triage["Triage"]
  Triage --> Class["Title and tags"]
  Class --> Index["Search index"]
  Index --> Confirm["Sample retrieval"]