Cortex: Allowing users to migrate their models from Ollama

[eckartal]

Overview

Users use Ollama to download, run, and store AI models. We should make this process smoother eliminating the need for re-downloading.

For inspiration: Email marketing tools provide amazing migration features to make migration as easy as possible.

Goals

• Create an intuitive migration interface inspired by top email marketing platforms such as ConvertKit
• Enable automatic detection and transfer of Ollama models to Cortex
• Maintain model settings and configurations during migration
• Minimize downtime and data transfer during the migration process
• Provide clear progress indicators and success/failure notifications

Questions

1. How can we ensure compatibility between Ollama and Cortex model formats?
2. What's the most efficient way to transfer large model files without re-downloading?
3. How should we handle partial or interrupted migrations?
4. What level of customization should we offer in the migration process?
5. How can we match or exceed Ollama's speed in supporting new models post-migration so that users don't quit?