AI Provider Setup

Choose and configure your AI provider for timeline analysis

Dayflow supports three AI processing modes, each with different privacy, quality, and cost trade-offs.

Quick Comparison

Provider	Privacy	Quality	Cost	Speed	Requirements
Gemini	Cloud (BYO key)	Excellent	Pay-per-use	Fast (2 LLM calls)	API key
Local Models	Fully private	Good	Free	Slower (33+ calls)	Ollama/LM Studio
ChatGPT/Claude	Cloud	Best	Subscription	Medium (4-6 calls)	Paid subscription

Processing Efficiency

The number of LLM calls affects both speed and cost:

• Gemini: 2 LLM calls per batch (video upload + transcribe → generate cards)
• ChatGPT/Claude: 4-6 LLM calls per batch (extract frames → batch describe → merge → cards)
• Local Models: 33+ LLM calls per batch (30 frame descriptions + merge + title + merge check + merge cards)

Gemini’s native video understanding makes it the most efficient option. Local models require describing each frame individually, resulting in more API calls and slower processing.

Choose Your Provider

Select the provider that best matches your needs:

Gemini

Fast cloud processing with bring-your-own API key

Local Models

Complete privacy with on-device processing

ChatGPT/Claude

Premium quality with frontier reasoning models

Configuration Location

All AI provider settings are configured in:

Dayflow → Settings → AI Provider

You can switch providers at any time. Existing timeline data remains unchanged.