You do not need expensive subscriptions to build a capable personal AI assistant. A combination of free or low-cost tools can automate much of your daily digital workload. Here is a practical setup.
The Core: A Local LLM
Ollama is a free, open-source tool that runs large language models locally on your computer. Models like Llama 3.1 (Meta), Mistral, and Gemma 2 run on modern laptops with 16GB RAM or more. Once installed, you have an AI that runs entirely offline — no usage fees, no data leaving your machine, no rate limits. The quality is below Claude or GPT-4o but entirely usable for many tasks: summarising documents, answering questions, drafting text.
Interface: Open WebUI
Open WebUI is a free browser-based interface for Ollama that looks and feels like ChatGPT. Install it via Docker and access it at localhost:3000. You can upload documents, switch between models, and maintain conversation history. It also supports connecting to OpenAI-compatible APIs if you want to add paid models alongside local ones.
Automation: n8n Free Tier
n8n’s free self-hosted tier connects your local AI to external services. A practical automation: trigger on schedule → fetch your emails (IMAP) → summarise with Ollama → send digest to Telegram. No ongoing cost once set up on a home server or old laptop.
Document Search: Private Document AI
AnythingLLM and PrivateGPT are open-source tools that let you upload PDFs, documents, and notes and then query them with a local AI — “what does this contract say about termination clauses?” — without sending the documents to any external service. The response quality depends on your local model quality but works well for summarisation and specific fact extraction.
What Paid Services Still Do Better
Claude, GPT-4o, and Gemini still significantly outperform local models for: complex reasoning, code generation for non-trivial problems, and nuanced writing. Local AI is best for routine automation and privacy-sensitive document processing; cloud AI for tasks requiring the highest capability.
