OpenSlides v1.1.0

What's New

Per-Session Workspace

Upload images and documents to each session's workspace. The AI references them when generating slides in Solo mode — OCR (via
Tesseract.js) extracts text from images so the agent understands visual content without requiring vision-capable models.

Solo Mode as Default

Solo mode (free-form HTML/CSS slides) is now the default generation mode. Each slide is generated as a fully self-contained HTML
document with complete visual freedom.

Model Auto-Discovery

In Settings, click Discover next to the model field to automatically fetch the list of available models from your configured API
endpoint. Works with OpenAI, Anthropic, and any OpenAI-compatible proxy (e.g. LiteLLM).

AI-Generated Session Titles

Sessions are now named after their content. After generation completes, the AI summarises the slide titles into a short session name
instead of using the first user message.

Codex Desktop Pet 🐾

A Petdex-powered desktop pet now lives in the preview area. Browse and search 1400+ sprites directly in Settings, pick one, and it
will wander around your workspace — running while slides are being generated, waving when they finish. Drag it anywhere; position is
remembered across sessions.

Bug Fixes

• Fixed white screen on startup caused by missing getSettings in the context bridge
• Fixed await outside async function (SyntaxError) in IPC handlers caused by a broken edit

Settings UI

• Settings modal is now a two-column layout: AI configuration on the left, pet gallery on the right
• Anthropic model discovery now correctly sends the required anthropic-beta: models-2023-06-01 header

---

OpenSlides v1.0.0

AI-powered presentation editor built with Electron + React.

✨ Features

AI Generation

• Describe your presentation in natural language — the AI generates a full deck slide by slide
• Template mode — AI fills structured layouts (title, content, two-column, quote, section, blank) with rich elements: headings,
  bullets, stats, cards, diagrams, images
• Solo mode — AI designs each slide as free-form HTML/CSS with full visual freedom; consistent theme (colors, fonts, style) applied
  across the deck
• AI-driven intent detection — the model decides whether to generate a new deck or edit existing slides
• Stop generation at any time with the abort button

Slide Editing

• Live preview rendered in a sandboxed iframe
• Manual slide editor: layout, background color, and individual elements
• Undo / redo (Ctrl+Z / Ctrl+Y)
• Drag-to-reorder slides
• Per-slide color themes and transitions (slide, fade, zoom, none)
• Brand logo overlay — configurable position, size, and opacity

Export & Sessions

• HTML export — self-contained single file, runs in any browser, includes navigation controls
• Multiple presentations saved locally, auto-restored on relaunch

🤖 Supported LLM Providers

• OpenAI — GPT-5, GPT-5.4 etc.
• Anthropic — Claude 4.6 Sonnet, Claude 4.6 opus, etc.
• Any OpenAI-compatible endpoint — Azure, Groq, Ollama, MiniMax, LiteLLM, etc.