Knowledge Assistant
Employees ask questions in natural language — the AI searches manuals, policies, and process documents and provides the relevant passage with citations.
Most Common Entry Point
ConsultingServices.aiAI Consulting for SMEsSolution in Detail
Corporate LLM: Unlocking Internal Company Knowledge with AI.
An internal AI assistant that understands manuals, process documents, and policies — providing your employees with the right answer in seconds. No more hours spent searching through SharePoint folders.
⏳ Time-to-Value
4–8 weeks
Investment (One-time)
from €6,500 (Basic)
from €14,500 (Pro)
Ongoing Costs
approx. €100–500 / month
(Azure Hosting & LLM APIs)
Deliverables
Basic: Secure chat UI, Azure OpenAI backend, SSO
Pro: + RAG connection to internal company documents (SharePoint/Teams)
Target Group
Who is this for?
Fits well if …
- Knowledge is distributed across dozens of PDFs, wiki pages, SharePoint folders, and email threads
- New employees take weeks to find relevant processes
- The same follow-up questions keep going to the same people
- Internal policies exist, but no one knows exactly where
- You have 15+ employees accessing shared knowledge
Less suitable if …
- Your company does not have documented processes yet
- Fewer than 50 documents exist as a knowledge base
- Knowledge is passed on purely verbally and cannot be documented
Application Areas
Where Internal LLM Has the Greatest Leverage
Three common scenarios where a Corporate LLM delivers immediate tangible benefits.
Onboarding AI
New employees ask questions about the company, processes, and tools — and receive immediate context-based answers instead of asking colleagues.
Saves 40% onboarding timeCompliance Assistant
Checks texts, offers, or contracts against internal policies and highlights potential deviations. Not a substitute for legal advice — but a strong first filter.
For regulated industriesBenefits
What Changes Specifically
Instead of 15 minutes in the wiki or SharePoint: precise answers in under 10 seconds.
New employees get oriented faster — without constantly asking colleagues.
Every answer shows exactly which document it comes from — verifiable.
Data does not leave your infrastructure. On-premise or EU cloud operation possible.
Model calculations based on typical SME scenarios. Individual results may vary.
Process
This is How Your Corporate LLM is Created
Mapping Knowledge Sources
Together we identify all relevant sources: documents, wikis, email archives, databases. We clarify access rights and data formats.
→ Source inventory + permission matrixIndexing & Preparation
Documents are parsed, divided into meaningful sections, converted into vectors, and indexed. OCR for scanned PDFs included.
→ Indexed knowledge base + quality report (coverage, gaps)Pilot & Fine-tuning
The assistant is tested in a small group. Answers are evaluated, thresholds adjusted, escalation paths defined.
→ Pilot system with access for test group + evaluation reportRollout & Knowledge Maintenance
Expansion to all employees, integration into existing tools (Teams, Slack, Intranet). Regular updates of the knowledge base.
→ Productive system + maintenance concept + usage statistics after 4 weeksArchitecture Decision
Small Language Models (SLMs) vs. Cloud AI
There is no "best model". Huge cloud models (like GPT-4) are great for complex tasks. However, for many everyday workflows in SMEs, so-called "Small Language Models" (SLMs) and Edge AI are often the smarter choice. We advise you independently.
Proprietary Cloud Models
(e.g. GPT-4o, Claude 3.5, Gemini)
The standard for quick results.
- Highest performance & precision immediately usable
- Easy cloud connection without own servers
- Dependency on the provider and ongoing API costs
Small Language Models (SLMs) & Edge AI
(e.g. Llama 3 8B, Phi-3)
Efficient, cost-effective, and completely data protection compliant.
- Full data control: On-premise or edge operation
- Nearly zero ongoing API costs (one-time setup)
- Perfect for industrial plants or highly sensitive law firm software
- Requires own hardware (GPUs) and setup know-how
How we stop data hallucinations: Whether SLM model at the edge or cloud GPT-4 – through our methodology ("Retrieval-Augmented Generation") we prohibit the models from guessing. They strictly cite only from your uploaded documents.
Under the Hood
Technical Setup
This is how the architecture is built — transparent instead of a black box.
Document Ingestion Pipeline
PDFs, DOCX, HTML, Confluence pages, and emails are automatically parsed. OCR processes scanned documents. Metadata (author, date, department) is included.
Chunking & Embedding
Documents are semantically divided into sections (not by character count, but by units of meaning). Each chunk is stored as a vector — enabling search by meaning, not just keywords.
Retrieval-Augmented Generation
When a question is asked, the most relevant document chunks are retrieved and provided to the LLM as context. The model generates the answer solely based on these sources — no hallucination.
Access Control (RBAC)
Not everyone can see everything. Role-based access rights ensure that the assistant only returns documents that the querying user is allowed to view.
Guardrails & Prompt-Hardening
System prompts are hardened against injection attacks. Output filters prevent the disclosure of confidential data outside the allowed context. Answers in case of uncertainty: "I don't know".
Audit Log & Monitoring
Every request is logged: Who asked what and when? Which sources were cited? Dashboard with usage statistics and unanswered questions feed.
Typical Stack
The stack is tailored to your requirements for data protection and integration. Fully on-premise possible with open-source models (Llama 3, Mistral). Azure, AWS, or your own servers — you decide.
Data Protection & Compliance
GDPR Compliance is Not an Add-On Feature
Your documents do not leave your infrastructure. On-premise deployment or EU cloud (Azure/AWS Frankfurt) — you choose.
Your company data does not flow into the training of external models. API calls are contractually excluded from training.
Complete audit log of all requests and responses. Deletion concept and retention periods configurable according to your data protection officer.
The assistant supports decisions — but does not make them automatically. Human control remains intact.
Frequently Asked Questions
Corporate LLM — Answered Specifically
How do the service packages differ?
In the Basic Package, you receive the fast, immediately deployable standard solution: Secure chat UI, Azure OpenAI backend, SSO. Ideal for easily proving value. The Pro Package is intended for deep system integrations: + RAG connection to internal company documents (SharePoint/Teams). Here we place special emphasis on enterprise readiness, customizing, and scaling.
Do I need to prepare all documents in advance?
No. The ingestion pipeline automatically processes PDFs, Word files, HTML, and scanned documents. What I need: Access to the sources and a brief overview of which areas should be covered.
Can employees have different access rights?
Yes. Access control is based on your existing roles (e.g. Azure AD / Entra ID). The assistant only shows answers based on documents that the user is allowed to see.
Does the system run in the cloud or locally?
Both are possible. Cloud: Azure or AWS (EU data centers). On-premise: own server with open-source models (Llama 3, Mistral). Hybrid forms are also possible — e.g. vector database locally, LLM via Azure API.
How current are the answers?
The knowledge base is regularly updated — automatically when changes occur in connected sources or manually via re-indexing. New documents are available within minutes, depending on the setup.
How much does a Corporate LLM cost?
Typically in the Professional Package from €6,900. On-premise setups with hardware consulting in the Enterprise Package. Ongoing costs: €50–300/month for hosting and API depending on usage volume.
Corporate LLM from the Professional Package — with deliverables.
Voice AgentsAutomate telephone availability.
ChatbotsIntercept standard questions on website and email.
Self-Assessment
Do We Need Our Own Corporate LLM?
Answer these 5 short guiding questions and receive an immediate assessment of how much potential this service holds for you.
Question 1 of 5
Do your employees potentially use sensitive data with public AI tools (like ChatGPT)?
Next Step
We will clarify whether your internal knowledge is ready for an LLM in 45 minutes — free and without obligation.
Request AI Initial Consultation nowConcrete Offer
What you get, how long it takes, and how risk is reduced.
- Result
- Internal AI chat with sources, role concept, answer boundaries, and quality test with real employee questions.
- Timeframe
- 4-8 weeks
- Price anchor
- from 9,900 EUR
- Best fit
- Best when knowledge is scattered across documents, SharePoint, email, and people.
Risk reduction
- Pilot before rollout
- Human-in-the-loop and fallback rules
- Documented data flow and handover
Proof material
Review sample deliverables before deciding: pilot report, implementation plan, prompt and fallback set, handover documentation.
View work examplesStandard process
- Maturity check and initial consultation
- Scoped pilot with realistic data
- Rollout decision and handover
Not included by default
External licenses, large-scale data cleanup, major ERP/CRM rebuilds, and legal case-by-case advice are scoped separately before project start.