Internal Knowledge Assistants That Show Where an Answer Comes From[1]
Help employees find answers in approved SOPs, policies and support documents.
Discuss an internal knowledge-assistant pilot with MTI Tech. Use retrieval-augmented generation, or RAG, to connect questions with relevant company information, show supporting sources and respect each user’s access permissions.
Refund requests over the set limit need a manager's approval.[1]
Find relevant sources before answering.
Retrieval-augmented generation combines search with AI-generated answers. It retrieves relevant passages from an approved collection and uses them to support a response, helping employees find information without searching every document manually.
Answers and citations still need evaluation. A source link does not guarantee that an answer is correct.
A manager approves refunds above the set limit.[1] Requests are logged in the support queue.[2]
Start with an approved document collection.
Choose a useful collection with a named owner, current documents and a clear update process.
Confirm which sources are authoritative before indexing them.
Resolve outdated and contradictory documents before relying on them for answers.
Five controls behind useful answers.
Sources and ownership
Identify approved documents and the people responsible for maintaining them.
Access
Limit retrieval to the documents each user is allowed to see. Test those permissions before rollout.
- Refund procedure
- Escalation contacts
- Approval limits by manager Manager only
Freshness
Track versions and update the index when source documents change or become obsolete.
Evidence
Show relevant source passages so users can inspect the basis of an answer.
Manager approval is required above the limit.[1]
“…requests above the set limit require approval from a manager before…”
Fallback
Acknowledge missing or conflicting evidence and direct the user to the document owner.
I couldn't find enough approved material to answer this. The responsible person is the policy owner in HR Operations.
Test retrieval, answers and permissions.
Review common, ambiguous and unanswerable questions. Include access-sensitive questions and recently changed policies. Assess retrieved sources, answer support, citations and escalation separately.
Evaluation identifies where the assistant can help and which questions still need a person.
Keep your knowledge assistant current.
Confirm document ownership and access before adding more users or departments. Scope and cost depend on source volume, permissions, integrations and evaluation needs.
Before you index anything.
Not necessarily. RAG retrieves relevant source material for an answer. Compare it with existing search and knowledge tools before deciding whether a more complex architecture is useful.
Access must be scoped. Start with approved sources and defined users, then test permissions before adding departments or document collections. Private material should remain unavailable to unauthorized users.
Source priority, ownership and version rules determine which material is authoritative. Unresolved conflicts should be shown to the user or escalated to the document owner.
An internal assistant focuses on employee questions, company documents and role-based access. A public chatbot serves customer questions and needs a separate content scope and support handoff.
Which answers does your team keep searching for?
Bring recurring questions, their source documents and your access requirements. Start with one team and a clearly owned knowledge collection.