Based on LangChain framework
state-of-the-art

✅ - AI Chat App running on local network
✅ - File upload
✅ - Web Search
❌ - Indexing local files
❌ - RAG Orchestrator

• Install Ollama
• Pull some models to mess with

ollama pull deepseek-r1:14b # reasoning model
ollama pull mistral:latest # general purpose

• Install Docker

Here's a sample of the final result:

# Pull the latest image
docker pull ghcr.io/open-webui/open-webui:main

# Volume to persist data
docker volume create open-webui-data

# Run the container
docker run -d -p 6969:8080 \
	-v open-webui-data:/app/backend/data \
	--name open-webui \
	ghcr.io/open-webui/open-webui:main

# Check container status
docker ps

# Get your IP
ipconfig getifaddr en0

Visit http://<your_ip>:6969 (or http://localhost:6969) in your browser.

Note:
On first run, you will be prompted to create an admin account. Other user accounts can be created later on.

Go to Settings > General > System Prompt and set it to your liking.

You are the most efficient AI Assistant that answers following these principles:

- Casual, straight-to-the-point responses.
- Prioritise IT best practices and performance.
- Provide trade-offs when applicable.
- Fact-check and provide sources when needed.

Go to Settings > Advanced and set the following parameters:
so it knows how many R are in strawberry

Mirostat: 0 # no randomness
Top K: 10 # choose within the top 10 tokens
Frequency Penalty: 1 # no recall/repeat
Max tokens (num_predict): 4096 # to be safe

Note:
We are using Google Search Engine here, but other search engines are available. Full documentation is available here

1. Create a Google Custom Search Engine
2. Grab the Search Engine ID created
3. Get the API Key from Custom Search JSON API

Go to Settings > Admin settings > Web Search
Set the Web Search Engine to be google_pse and fill in the API Key and Search Engine ID.

Search Result Count: 5
Concurrent Requests: 6