AI guide

What is RAG in generative AI?

RAG means the model is handed excerpts from your own material before it answers. It is how you stop a chatbot inventing policy. It only works if the right excerpt is found and the person asking is allowed to see it.

Designed in Dubai and Ajman.

Dubai office

45th Floor, The One Tower, Sheikh Zayed Road, Barsha Heights

Ajman office

26th Floor, Amber Gem Tower, Nuventure Free Zone

Contact

Email, call, or WhatsApp info@raabsoftwares.com

Call
DubaiThe One Tower, Barsha HeightsAjmanAmber Gem TowerOffices

What we build

  • The moving parts
  • Where it fails
  • A sensible first version
What is RAG in generative AI? | Raab Group Software Solutions UAE

AI guide

The moving parts

Vectors are a common way to search by meaning. Filters on department, date, and document type still do a lot of the work. A RAG system that is “just a vector database” will mix last year’s policy with this year’s.

The moving parts - What is RAG | Raab Group Software Solutions UAE

AI guide

The moving parts

How we work

    The moving partsDocuments are split into passages and indexed. A question is used to retrieve a handful of passages. Those passages are placed in the prompt, and the model is told to answer from them and to say when they are not enough. The index is not the product. The product is the answer a specific person was allowed to receive, with a way back to the source.
    Where it failsThe right paragraph was never retrieved, so the model fills the gap. Two documents disagree and the model blends them. A scanned PDF was indexed as garbage. A user retrieves a file their colleague should not have shared. None of these are fixed by a more confident prompt. They are fixed by the corpus, the permissions, and a test set of real questions.
    A sensible first versionOne collection with an owner. Permissions copied from the source. Citations on every answer. A refusal when retrieval is weak. A pilot group and a list of failures. The knowledge assistant sample shows that shape. It is a concept, not a client claim. The RAG development page is how we deliver it.
    What Is RAG in Generative AI? | Raab Group | Raab Group Software Solutions UAE
    What is RAG

    In practice

    Retrieval is permissions plus sources

    A plain explanation of retrieval-augmented generation.

    Documents are split into passages and indexed. A question is used to retrieve a handful of passages. Those passages are placed in the prompt, and the model is told to answer from them and to say when they are not enough. The index is not the product. The product is the answer a specific person was allowed to receive, with a way back to the source.

    View this work
    • Search over documents you control
    • An answer with something the user can open
    • A refusal when the source is not there
    • The model memorising your company
    • A licence to ignore access control
    • A guarantee of truth

    On the screen

    Where it fails

    The right paragraph was never retrieved, so the model fills the gap. Two documents disagree and the model blends them. A scanned PDF was indexed as garbage. A user retrieves a file their colleague should not have shared. None of these are fixed by a more confident prompt. They are fixed by the corpus, the permissions, and a test set of real questions.

    Questions we are often asked

    Where it fails - What is RAG | Raab Group Software Solutions UAE

    What is RAG

    • The moving parts
    • Where it fails
    • A sensible first version

    In this build

    What is RAG

    What Is RAG in Generative AI? | Raab Group | Raab Group Software Solutions UAE
    What is RAG
    • The moving partsDocuments are split into passages and indexed. A question is used to retrieve a handful of passages. Those passages are placed in the prompt, and the model is told to answer from them and to say when they are not enough. The index is not the product. The product is the answer a specific person was allowed to receive, with a way back to the source.
    • Where it failsThe right paragraph was never retrieved, so the model fills the gap. Two documents disagree and the model blends them. A scanned PDF was indexed as garbage. A user retrieves a file their colleague should not have shared. None of these are fixed by a more confident prompt. They are fixed by the corpus, the permissions, and a test set of real questions.
    • A sensible first versionOne collection with an owner. Permissions copied from the source. Citations on every answer. A refusal when retrieval is weak. A pilot group and a list of failures. The knowledge assistant sample shows that shape. It is a concept, not a client claim. The RAG development page is how we deliver it.

    Where it fails

    The right paragraph was never retrieved, so the model fills the gap. Two documents disagree and the model blends them. A scanned PDF was indexed as garbage. A user retrieves a file their colleague should not have shared. None of these are fixed by a more confident prompt. They are fixed by the corpus, the permissions, and a test set of real questions.

    Fine-tuning does not replace this. Fine-tuning changes style or behaviour. It does not keep a living policy library up to date. Most companies should spend the first budget on retrieval and evaluation.

    • The model memorising your company
    • A licence to ignore access control
    • A guarantee of truth
    Where it fails - What is RAG | Raab Group Software Solutions UAE
    Where it fails

    A sensible first version

    One collection with an owner. Permissions copied from the source. Citations on every answer. A refusal when retrieval is weak. A pilot group and a list of failures. The knowledge assistant sample shows that shape. It is a concept, not a client claim. The RAG development page is how we deliver it.

    A sensible first version - What is RAG | Raab Group Software Solutions UAE
    A sensible first version

    Included now

    Retrieval is permissions plus sources

    RAG is

    • Search over documents you control
    • An answer with something the user can open
    • A refusal when the source is not there

    RAG is not

    • The model memorising your company
    • A licence to ignore access control
    • A guarantee of truth

    Decide before we start

    Questions that define a RAG project

    • W
      Which filesCurrent policy is in. Personal drives are a decision.
    • W
      Which userThe index has to respect the same permissions as the files.
    • W
      Which updateA replaced document has to leave the answers. Say who checks.

    Bring this to the call

    Test the idea against your files

    A

    A current policy

    One document the answer should cite.

    A

    A retired policy

    One document the answer must not use.

    A

    A role

    A person who must not see the first document.

    A

    A refusal

    A question with no source, which should not be invented.

    Related

    Related reading

    Common questions

    Questions we are often asked

    Is RAG the same as search?

    Search returns documents. RAG returns an answer composed from retrieved passages, and it can still be wrong. Keep the citation so a person can check. For some tasks, ordinary search is the better product.

    Can RAG use our CRM as well as files?

    Yes, if you treat records as documents with permissions. A customer-facing bot and a staff assistant should not share one unrestricted index.

    Does the model store our documents?

    Your index stores the passages. Whether a model provider also retains prompts is a contract and configuration question. We do not send material to a service your policy forbids, and we say which service is in the design.

    Published 2026-06-10 · 8 min read

    Ready to operate

    Marks this RAG explainer stays precise

    1. 01
      A source is requiredThe article says an answer should point at a document.
    2. 02
      Missing is an answer“Not in the files” is described as a valid result.
    3. 03
      Access follows filesRetrieval does not bypass who may open the document.
    4. 04
      Not a chatbot sloganThe piece separates retrieval from a bot that only chats.

    Use this list on the call. Tell us which marks are missing today, and which can wait.

    Two offices

    Dubai and Ajman.

    The One Tower in Barsha Heights, and Amber Gem Tower in Ajman.

    Offices

    Write to us

    Tell us about the work.

    Share the outcome, the deadline, and the constraints. We reply from Dubai or Ajman with a practical way to start, including what should wait.

    Dubai 45th Floor, The One Tower, Sheikh Zayed Road, Barsha Heights
    Ajman 26th Floor, Amber Gem Tower, Nuventure Free Zone