Building a Personal Knowledge Base with Local Documents in Open WebUI

A personal knowledge base is the single most useful local AI application. Instead of searching through scattered notes, PDFs, and bookmarks, you upload everything to Open WebUI and ask questions in natural language.

This guide walks through the entire process: organising your source documents, importing them, configuring retrieval, and building a daily workflow.

What Makes a Good Personal Knowledge Base

| Document type | Include? | Why |

|--------------|----------|-----|

| Personal notes | Yes | Core of your knowledge base |

| Technical docs | Yes | Frequently referenced |

| Research papers | Yes | Ask questions across papers |

| Book highlights | Yes | Retrieve by topic |

| Web clippings | Yes | Capture ephemeral content |

| Work documents with PII | No | Security concerns |

For security considerations, read How to Index Local Documents Safely on a Private Server.

Setting Up Open WebUI

```bash

docker run -d \

-p 3000:8080 \

-e RAG_EMBEDDING_MODEL=nomic-embed-text:v1.5 \

-v open-webui-data:/app/backend/data \

ghcr.io/open-webui/open-webui:main

```

Pull your embedding model:

```bash

ollama pull nomic-embed-text:v1.5

```

Create a "Personal Knowledge Base" workspace to keep personal documents separate.

Preparing Your Documents

Clean up formatting

• **PDFs:** Ensure text-based, not scanned images
• **Markdown:** Standard markdown works well
• **Plain text:** Remove excessive line breaks

Organise by topic

```

knowledge-base/

├── personal/

├── technical/

└── research/

```

Configuring Retrieval

Adjust chunk settings for personal notes:

```bash

-e RAG_CHUNK_SIZE=300

-e RAG_CHUNK_OVERLAP=50

-e RAG_TOP_K=5

-e RAG_RELEVANCE_THRESHOLD=0.6

```

Enable hybrid search for better precision on names, dates, and project titles.

Building a Daily Workflow

Ask your knowledge base:

• "What were my priorities yesterday?"
• "Summarise key decisions from last week's project notes."
• "What books am I currently reading?"

Save new article highlights immediately as text files and upload them.

Backing Up

```bash

docker stop open-webui

cp /path/to/data/webui.db /backup/personal-kb-$(date +%Y%m%d).db

docker start open-webui

```

Conclusion

Start small with your current notes and a handful of reference documents, then add content as you identify gaps. The most valuable knowledge bases are built gradually.

FAQ

How many documents can Open WebUI handle?

Thousands comfortably. Performance depends on your embedding model and available memory.

Can I share my knowledge base?

Documents in a workspace are only visible to workspace members.

Does it work offline?

Yes. Everything runs locally — no internet access required.

**Sources:**

• Open WebUI GitHub Repository
• How to Index Local Documents Safely on a Private Server
• Build a Local RAG Pipeline That Actually Answers Questions