Knowledge · Australian teams
Company knowledge AI and RAG
Help staff find answers in approved company information, with source references and permission checks.
Understand the service
What this means in practice
A company knowledge AI tool retrieves relevant information from an approved collection before a language model prepares an answer. This pattern is often called retrieval-augmented generation, or RAG. It is different from training a new model on all your documents: retrieval selects context for a particular question.
A useful answer needs the right source, the right version and the right access. Temrik’s product source includes portal chat and evidence-led Company Intelligence; availability depends on deployment, authorised sources and configuration. An engagement should begin by checking those dependencies, then testing a small knowledge collection against questions your staff actually ask.
- Microsoft: retrieval-augmented generation Retrieval supplies context; source preparation and evaluation matter.
- Microsoft: document-level access control Document permissions and their synchronisation require explicit design.
A practical workflow example
An employee asks which onboarding checklist applies to a new service. The proposed assistant retrieves the approved checklist, identifies its revision and links to the source. If two policies disagree, it surfaces the conflict for the document owner. If the employee lacks access, the restricted text must not enter the answer context.
Proposed engagement
How we would approach the work
Choose one collection
Start with an owned set of procedures or product guidance. Identify obsolete copies, scanned documents, access groups and update responsibilities.
Test retrieval and answers
Use answerable, unanswerable and access-restricted questions. Check the retrieved passages separately from the final wording and verify every citation supports the answer.
Operate the knowledge lifecycle
Agree refresh intervals, permission changes, deletion behaviour and a correction route. Retest when documents or the model change.
Deliverables to agree in the scope
- A source register with ownership, versions and access rules.
- A scoped retrieval design and reference question set.
- An evaluation report covering answer support, abstention and access boundaries.
Access, sample information and reviewer availability affect the plan. Any implementation, provider costs, support arrangements and acceptance criteria are agreed before work begins.
Limits worth understanding
- RAG does not guarantee factual answers or automatically preserve source-system permissions.
- Storage location, retention and model-provider terms must be verified for the selected deployment.
Questions to bring to the first conversation
- Which documents can this audience actually see?
- How quickly must a changed or deleted policy disappear?
- What should the assistant say when the evidence is incomplete?
Australian teams
Scope the work for your operating context.
For Australian teams, start with a clearly owned set of operating procedures and identify which apply nationally or only in a state. Review personal information in those documents before ingestion, and verify provider handling rather than assuming Australian hosting.
A starting reference for your review: OAIC: using commercially available AI products. Local obligations and deployment settings need to be assessed for the actual use case.
A focused next step
Work with Temrik.
Tell us about the workflow you want to improve and the outcome you need. We can review the context and discuss a focused assessment. Scope and price are agreed before paid work begins.