Aylian StudiosAylianStudios
Back to changelog
Aylian Studios

Raphael v2.3

releaseaidiscord

Vision Processing

  • Image Attachment Extraction: Discord bot now detects and extracts image attachment URLs from every message before processing.
  • AIClient::analyze_image(): Sends image URLs to a vision model — LocalAI (LLaVA) first, with OpenCode as fallback — and returns a natural language description.
  • Pipeline Integration: Image descriptions are injected alongside message text content, meaning wiki pages generated from image-heavy conversations now include accurate visual context.

Like what you read?

Subscribe to our newsletter to never miss an update.

Stay Updated

Get notified when we publish new changelogs, updates, and project releases.