Have AI chatbot developed

AI assistant with Your own company data

CodeGuides develops AI chatbots that access your own documents, knowledge bases and data sources, with configurable access rights, attribution and integration into your system environment.

  • RAG architecture: answers from our own documents and sources
  • Configurable access rights and data sharing
  • Integration into existing web apps, intranets and tools

Bring specific requirements and we will check the data situation and integration options.

  • RAG architecture with own data
  • Source per answer
  • Access rights configurable
  • Integration into your systems

This is what a productive AI chatbot with its own company data looks like

Connected sources

Manuals & Terms and Conditions
234 documents
Support Wiki
1,420 items
CRM customer groups
API connection

Access rights

Support Team: All sources
Sales: Manuals + CRM
Customers: Public FAQ

AI assistant Online
What return period applies to business customers?
According to the B2B General Terms and Conditions, an individual return policy applies to business customers. In the standard configuration, the deadline is 14 business days from the delivery date, unless otherwise agreed in the contract.
Sources (3)
Contract_B2B_AGB.pdf · p. 14
Support Wiki: Return Types & Deadlines
CRM customer group: Business customers (Premium)
Enter your question…

Selected projects and customers

Wildau University of Technology
Secanda AG
SENSYS GmbH
Knorr-Bremse AG
Zaibr Innovations GmbH
DTAD AG
Findeling GmbH
Netcom GmbH
Own data sources RAG architecture API integration Source based answers Individual development

Why simple chatbots not enough

An AI chatbot without a connection to your own data will not answer company-specific questions.

Standard FAQ bot

Predefined answers to known questions

Only fixed answers to pre-programmed questions
No connection to internal documents or systems
High maintenance effort for every change in content
No source information, no traceability

AI chatbot with RAG

Answers from our own documents and data sources

Understands free questions and looks for suitable answers in your sources
Connection to PDFs, wikis, databases and APIs
New documents = new answer basis, no manual effort
Every answer with reference to the source, transparent and verifiable
General AI doesn't know your content

ChatGPT and Co. do not know any internal manuals, product data or company-specific rules.

Free text search doesn't find the right thing

Full-text search returns hits, not answers, employees have to read and interpret themselves.

No source credit

Without citing the source, it is difficult to check whether the answer is correct, which slows down productive use.

Data protection and access rights are missing

Without a configured access concept, internal data can unintentionally become accessible to all users.

No system integration

An isolated chatbot is of little help, it has to be available where the work takes place.

Hallucinations without a database

AI without verified data base invents answers, RAG only answers what is in approved sources.

An AI chatbot that knows your own content and answers comprehensibly.

CodeGuides develops AI assistants with RAG architecture: answers based on your shared documents, with source information and configurable access rights.

Ask a question
Search sources
Answer with sources

This is how RAG works, Retrieval-Augmented Generation

RAG is the architecture that enables AI chatbots with their own company data. This is how a request works:

1

User question

User asks free question in the chatbot

2

Vector search

Semantic search in indexed sources

3

Recall passages

Relevant text sections from approved sources

4

LLM formulated

AI generates answer from retrieved passages

5

Answer + sources

Answer with sources, transparent and verifiable

Important: RAG does not mean hallucinationsThe language model only responds based on the retrieved passages from your shared sources. If the system doesn't find any suitable information, the chatbot will say so clearly instead of making up an answer.

Data sources

Which data sources can the AI chatbot use?

The AI ​​assistant connects to the data sources you share, structured, indexed and with configurable access rights.

Documents

PDFs, Word documents
Manuals & Instructions
Technical data sheets

Knowledge base

Confluence / Wiki
SharePoint / Intranet
Notion, Gitbook

Systems & APIs

CRM data (customer groups)
ERP product data
Support ticket systems

AI chatbot

RAG architecture
Access rights configurable

Web & Intranet

Company website
Intranet / employee portal
Customer portal

Software products

SaaS application
Web app / dashboard
API for external systems

Communication

Microsoft Teams
Slack
Email based inquiries

What AI chatbots are for can be used productively

Concrete use cases that are technically feasible and economically sensible.

01

Internal knowledge assistant

What are the vacation regulations for short-time work?
According to Section 3 Paragraph 2 of the works agreement, the vacation entitlement for short-time work is adjusted proportionally...
Company Agreement_2024.pdf · §3

Employees spend hours searching through manuals, guidelines and old project documents.

RAG-based AI assistant in the intranet with configured document sources, access rights per team and source citation.

  • Faster information search
  • Fewer internal queries
02

Product data assistant for sales

HP Series 200 · High pressure pump
Max. 220 bar · DN50 · IP67 · Data sheet: HP200_v3.pdf
HP Series 250 · High pressure pump
Max. 250 bar · DN65 · IP68 · Data sheet: HP250_v2.pdf

Sales teams manually search technical product information, specifications and compatibility in PDFs and internal systems.

AI assistant with access to product catalogs, data sheets and technical documentation, with precise answers and source citation.

  • Faster quotation processing
  • Fewer queries to the technical department
03

Customer Support Assistant

When will my order arrive?
Standard deliveries are delivered within 3 to 5 working days. Express delivery possible until 6 p.m. the next working day.
Support FAQ: Shipping & Delivery

Incoming customer inquiries are processed manually. Standard questions about products, delivery times and service tie up capacity.

AI assistant on the website or in the customer portal answers standard questions based on released product information and FAQ.

  • Less manual standard requests
  • 24/7 availability for common questions
04

Compliance and Policy Wizard

Does ISO 27001 apply to our cloud providers?
According to IT security guidelines §8, all external cloud services must provide evidence of ISO 27001 certification or an AVV must be available.
IT Security Guideline_v2025.pdf · §8

Employees regularly ask about internal guidelines, compliance requirements and legal requirements.

Internal chatbot with access to current guidelines, manuals and compliance documents, with versioned source references.

  • Faster answers to policy questions
  • Always current information base
05

Support Team Assistant

Suggested answer
Known issue in v3.1.2, patched in v3.1.3. Workaround: Delete the app and reinstall it. Release: Wednesday.

Support agents manually search for answers in documentation and rephrase standard answers each time.

Internal assistant provides contextual answer suggestions for agents based on released product data and historical tickets.

  • Faster ticket processing
  • More consistent response quality
06

Onboarding assistant for new employees

How do I request VPN access?
VPN access is requested via the IT self-service portal at intern.firma.de/it. Activation within 1 working day.
Onboarding Wiki: IT equipment

New employees have many questions about processes, tools and internal procedures, thereby tying up capacity in experienced teams.

AI assistant with access to onboarding documentation, FAQs and process guides, available 24/7.

  • Less onboarding queries
  • Faster onboarding

Do you have a specific use case? We check the data situation and technical feasibility.

This is how we develop yours AI chatbot

From the data situation to the productive chatbot, structured and without a long lead time.

1

Clarify use case & goal

Which questions should the chatbot answer? Who are the users? What is the concrete added value?

2

Check data sources

What documents and systems are in place? Evaluate quality, format and accessibility.

3

RAG concept & architecture

Specify embedding model, vector database, access concept, integration and data protection architecture.

4

Develop prototype

MVP with limited data range. You test with real questions and real users.

5

Test response quality

Systematic testing with real questions, optimize source coverage and answer accuracy.

6

Integrate & expand productively

Integration into your system environment. Monitoring, feedback loop and expansion of additional data sources.

Why CodeGuides for your AI chatbot

An AI chatbot is more than a language model. It needs clean backend infrastructure, data pipelines and access concepts.

Backend & data pipeline

Document import, indexing, embedding and vector database, we build the entire infrastructure behind the chatbot.

Access rights & role concept

Fine-grained access concepts: Which group of users sees which document sources, appropriate to your organizational structure.

System integration

The chatbot must be accessible where your employees work: intranet, web app, CRM or as an API.

Sources & transparency

Each answer shows which source it comes from. This creates trust and makes mistakes recognizable.

Data protection by design

Data storage possible in Germany or the EU. Data protection is taken into account architecturally from the start.

Monitoring & Quality Assurance

Logging, feedback mechanisms and response quality monitoring for productive operations.

An AI chatbot is not a plug-and-play product. The crucial difference lies in the clean data pipeline, well-thought-out access concept and integration into the actual working environment, which is exactly our strength as a software development partner.

Who is an individual one for AI chatbot useful?

An individual AI chatbot with its own data makes sense when generic AI tools cannot access your content.

Well suited

  • Companies with many internal documents, manuals or product data
  • Teams with frequently recurring information requests
  • Support teams working with extensive documentation
  • SaaS providers who want to integrate an assistant into their product
  • Companies with clear requirements for access rights and data protection

Less suitable

  • Companies without digitized document stocks
  • Very small amounts of documents that could also be mapped as a simple search
  • Without internal contact person for document maintenance and data maintenance

Unsure? In the initial consultation, we check whether and how a chatbot makes sense for your use case.

Ready to develop an AI chatbot with your own data?

We check your data situation, clarify access rights and integration options, and give an honest assessment of feasibility.

Data sources available?
Check use case
Plan chatbot MVP

Frequently asked questions about the AI chatbot

What is a RAG chatbot?

RAG stands for Retrieval-Augmented Generation. With RAG, the chatbot first searches shared documents and data sources and uses the information it finds as context for its response. This means it answers questions based on your own company content, not just based on general AI training.

How much does it cost to develop an AI chatbot?

A first MVP with RAG architecture and your own documents typically costs between 15,000 and 45,000 euros, depending on data volume, access rights complexity and system connection. In the initial consultation you will receive a concrete assessment.

Which documents can the AI ​​chatbot use?

PDFs, Word documents, Markdown, HTML pages, structured databases and systems connected via interfaces. We develop the document pipeline to suit your database.

Can an AI chatbot be implemented in a GDPR-compliant manner?

Yes. We develop with a data protection-oriented architecture: data storage in Germany or the EU, clear access rights, no sending of internal data to external services without a legal basis. We recommend additional legal examination.

Can the chatbot be integrated into our existing software?

Yes. We integrate the AI ​​assistant into web apps, intranets, CRM systems or provide it as an API. Integration is a central part of development.

What happens if the chatbot doesn't find an answer?

The chatbot clearly states that it cannot find a suitable source for this question in the shared documents instead of inventing an answer. Transparency is a core feature of the RAG approach.

How are access rights controlled?

We develop access concepts that define which user areas or teams have access to which document sources. No user sees more than what is approved.

Can we start and then expand?

Yes, we recommend an MVP with a limited data range. After proven operation, further sources, integrations and user areas can be added.

Do we need our own AI infrastructure?

No. If you don't want to build your own infrastructure, you can also use existing AI services (OpenAI, Azure OpenAI) and we will build the infrastructure around them. Vector database, document pipeline and backend are developed by us. Please note the data protection regulations.

Another question that isn't answered here? Discuss chatbot project
About CodeGuides

CodeGuides is an app and AI agency from Germany for Flutter apps, custom software, AI automation and local AI infrastructure. Consulting and implementation come from a single source: We evaluate use cases technically, build the first pilot and scale it up to company-wide operation, GDPR-compliant and with hosting in Germany.

CodeGuides GmbH · Königs Wusterhausen On the market since 2019 100+ digital projects 100% in-house (DE) Consulting & implementation from a single source

Request AI chatbot

Describe your use case. In the initial consultation we check the data situation and feasibility.

We will respond within one business day.

Or directly: info@codeguides.de · 03375 2510 343

Arrange a meeting

45 minutes, technical assessment of your chatbot project.

Franz Opitz, CodeGuides GmbH
Franz Opitz CodeGuides GmbH +49 (0) 3375 2510 343 info@codeguides.de

Was passiert danach?

1

Initial consultation (45 min.)

Clarify projects, data sources and integration options.

2

Technical assessment

RAG architecture, effort and concrete MVP suggestion.

3

Offer & implementation

Written concept with effort estimate. You decide.

Chatbot project check

This is what you should bring with you to the interview:

  • Which questions should the chatbot answer?
  • What data sources and documents are there?
  • Who are the users (employees / customers)?
  • Where should the chatbot be integrated?
Choose an appointment in the calendar