RAG proof of value: make your own documents searchable with AI and citations
Before an internal AI assistant is built, it should be clear whether it works with your documents. In the proof of value we make a scoped collection searchable with AI and evaluate the answers against agreed reference questions with verifiable sources.
Companies worldwide trust WZ-IT
Starting point
Your documents exist, but nobody finds the answers quickly
- Staff search file shares, wikis or manuals and end up asking colleagues anyway
- A department can name typical questions and the passages where the answer is found
- Before a company-wide assistant, you want to know whether it works with your documents
- Data location, access and citations must remain traceable
Evidence of quality for the production decision
Your department has worked with its own content and knows how good the answers are. The decision on an internal AI assistant rests on measured results rather than a demo.
Usable knowledge search
Employees ask natural-language questions and receive answers with traceable source references from the agreed content.
Defined access
We document who may use the knowledge area, how it is updated and which permissions must remain intact as it expands.
Measurable decision
Subject-matter tests, known limits and a clear recommendation show whether and how additional sources should be connected.
Knowledge still lives in conversations and experience? Our knowledge transfer pilot processes your interviews and documents through a pipeline your team can keep supplying with new content.
From priority content to tested everyday use
The technical implementation follows the business questions. We do not build a stack first and look for a use case afterwards.
- 01
Define scope
Confirm the knowledge domain, volume and formats. Define 40–60 reference questions for the PoV, or 60–100 test cases for the Connected Pilot.
- 02
Set up
We configure indexing, retrieval, a local or approved model and source citations.
- 03
Test
Results and answers are evaluated with the department and refined where required.
- 04
Handover
Outcome, limits, operational path and the next expansion stage are documented and handed over.
What we need from you
- A subject-matter contact for scope and acceptance
- Access to the agreed content or a representative export
- Subject-matter input for 40–60 reference questions, or 60–100 test cases for the Connected Pilot
- A short assessment of sensitivity, user groups and updates
One bounded source, completed through acceptance
One knowledge domain with agreed volume, reference questions with source and unknown-answer tests, two optimisation cycles and a quality report.
One knowledge domain with agreed volume and document formats
40–60 reference questions
source and unknown-answer tests
Baseline measurement and two optimisation cycles
Quality report and recommendation for production
Deliberate boundary
Not included in the proof of value
- Live connectors, organisation-wide SSO and complex permission inheritance
- Production operations, high availability and new GPU infrastructure
Scope and investment
Cost: RAG Proof of Value and Connected RAG Pilot
Two entry points: the proof of value tests a supplied collection. The Connected RAG Pilot additionally connects a live source with sign-in and permissions in a test environment.
RAG Proof of Value
€9,900
excluding VAT and separately quoted third-party costs
Credited against the production knowledge assistant. You receive the proposal for it with the quality report.
Evaluate a scoped document collection with reference questions and verifiable sources.
2–3 weeks
- One knowledge domain with agreed volume and document formats
- 40–60 reference questions including source and unknown-answer tests
- Baseline measurement and two optimisation cycles
- Quality report and recommendation for production
Outcome and verification: Baseline-to-final comparison, documented limitations and an evidenced go/no-go recommendation.
Connected RAG Pilot
€17,500
excluding VAT and separately quoted third-party costs
Credited against the production knowledge assistant. You receive the proposal for it with the quality report.
Validate one accessible standard source with synchronisation and a bounded access model in staging.
3–5 weeks
- One approved standard source with change and deletion synchronisation
- One SSO integration and an agreed basic RBAC profile
- 60 –100 tests covering answer quality, sources, updates and access
- Staging, two optimisation cycles, result acceptance and a production plan
Outcome and verification: Documented quality and access tests with a costable plan for production delivery.
Quoted separately: Custom connectors, IAM migration and complex ACL inheritance · Production SLA, high availability and organisation-wide rollout
Delivery starts once agreed data, access and stakeholders are available. The binding scope and price are defined in your proposal. Hardware, hosting, model consumption, licences and ongoing operations are quoted separately. Follow-on projects are costed against reusable results. We credit the pilot price when you commission the production rollout; there is no requirement to purchase every stage. A pilot concludes with a quality report and an evidenced recommendation to proceed or stop. A documented no-go is a valid outcome. Production approval and ongoing operations are agreed separately.
All prices are net and exclude statutory VAT. The offers are addressed to businesses.
From a tested knowledge area to a complete AI solution
The proof of value is the entry point to knowledge systems: next comes the internal AI assistant for everyday work. The tested content is reused.
Internal AI Assistant
Multiple knowledge areas, roles and assistants for everyday employee use.
View offerfrom €6,490 excl. VAT one-timeAI Cube
Local AI server in your own network, keeping knowledge area, index and model in-house.
Configure AI CubeCustom RAG and data integration
Custom connectors, ACLs, workflows, evaluation and interfaces for complex data landscapes.
View serviceFrequently asked questions about the RAG Proof of Value
Scope, data access, permissions and next steps
The RAG Proof of Value has a fixed content set, defined test questions, a clear deliverable and a fixed timeframe. Open-ended RAG consulting is appropriate when the data landscape, target design or integrations first need broader clarification. For complex starting points, we therefore define the right entry scope before work begins.
File upload provides a technical feature. The proof of value also addresses data quality, updates, permissions, test questions, citations and acceptance criteria for a defined company knowledge set.
No. The proof of value can run on an existing AI system, a suitable customer environment or as part of an AI Cube introduction. We agree in advance where data, index and model are processed.
Typical starting points include supplied file sets, network drives, Nextcloud, SharePoint exports, wikis or document management systems. Whether automated synchronisation fits the fixed price depends on the interface and permissions model.
That must be checked for each source. A simple pilot can use a bounded knowledge area. User-specific ACLs require a consistent identity and permissions path across source, index and interface.
We agree representative subject-matter questions and expected source passages. Retrieval, citation quality, completeness and recognisable edge cases are evaluated and documented rather than only shown in a demo.
The entry point is predictable: the RAG Proof of Value costs €9,900 excl. VAT, the Connected RAG Pilot with live source and permissions €17,500 excl. VAT. A production internal AI assistant starts at from €34,900 excl. VAT. Hardware, hosting and ongoing operations are quoted separately.
You can keep using the tested area. The usual next step is the internal AI assistant with further sources, roles and integrations; it is delivered only when the outcome and priority justify it. With a clear scope, a direct start without a proof of value is also possible.
Request a RAG Proof of Value
Which knowledge source should become usable first?
Tell us the source, user group and most important search or answer problem. We will check which package covers the right first step.
Let's Talk About Your Idea
Whether a specific IT challenge or just an idea - we look forward to the exchange. In a brief conversation, we'll evaluate together if and how your project fits with WZ-IT.
Arrange a callback
Companies worldwide trust WZ-IT
