What this covers
Search over your approved documents that respects permissions, links to sources, and says when it has no answer.
Tools
Embeddings, Vector search, Metadata filters, Document ingestion. The exact stack depends on what you already use.
Where it helps
- Searching internal product documentation
- Finding the right procedure
- Answering onboarding questions with links to sources
We agree what “done” means before work starts.
What you get
- Search across your approved documents, with source links
- Permission checks and a clear “no answer” response
- Source code, setup notes and an agreed definition of done
The final list depends on the agreed scope. The estimate and schedule follow from it.
Data and security
Access is filtered before search, document permissions are kept, deleted or outdated files are handled, and we test that one user can’t see another’s documents.
How I work on it
List the sources, set permissions, write a set of test questions, compare search approaches, then check answers and refusals.
What helps me get started
- A short description of the process and who uses it.
- Sample data, any API documents, and what access is available.
- How much it will be used, who reviews the work, and your release priorities.
A short outline is enough for a first call. Please send sample data, not real records or passwords.
Related services
Read the planning guideQuestions
Does RAG stop the AI from making things up?
No. It gives the AI better context, but answers still need testing, source links, and a way to say “I don’t know”.
How do you price the work?
I look at the workflow, the systems involved, the access needed and what “done” means, then send an estimate. If there are big unknowns, we start with a short discovery or a prototype.