Zero-Config AI

ɳSelf gives every user unlimited AI in 30 seconds. No API keys to manage, no billing to configure, no provider accounts to create. Run nself doctor --ai and you have a working AI stack.

Architecture

Zero-Config AI combines three tiers into a single, transparent system:

1. Local (Ollama) — Free, private, unlimited. Runs on your hardware. Best for embeddings, classification, and background tasks.
2. Gemini Pool — Free tier from Google. Each connected Google account adds 20 RPD. Pool multiple accounts for 100+ RPD at zero cost.
3. Paid APIs — OpenAI, Anthropic, or paid Gemini tiers. Used only when free tiers are exhausted or when quality requirements demand it.

How It Works

The AI routing engine evaluates each request and picks the best available tier based on task type, quality requirements, latency constraints, and remaining capacity. The goal: maximize quality while minimizing cost, defaulting to $0.

Task Classes

Key Principles

• Zero configuration required. Works out of the box.
• Local-first. Sensitive data stays on your machine.
• Cost approaches $0 for most workloads.
• Transparent. Every routing decision is logged.
• User controls everything. Override any default via Settings > AI.

Next Steps

• Quickstart — Get running in 30 seconds
• Local Models — Ollama setup and model management
• Gemini Pool — Connect Google accounts for free API access
• Routing — How the routing engine works
• Security — Encryption, OAuth, and audit