Roadmap

[nekomeowww]

v0.4

• UI
  • Logo design, Icon design
  • Provider options
    • Design
  • New sidebar nav design
  • Live2D from file (feat(stage-web): load and save live2d model #41) @LemonNekoGH
  • Better bundle for Electron @LemonNekoGH
  • Better drap and keyboard mapping settings for Electron @LemonNekoGH
  • Loading screen @LemonNekoGH @nekomeowww (more PR welcome)
  • Changable theme color (from @LemonNekoGH)
  • Tooltip when loading (from @nekomeowww)
  • Prompt user to set the API key and base URL
  • Error handling
  • Active providers for different modules
  • Module configuration (for https://github.com/moeru-ai/airi/pull/43/files#diff-ab31ccf7a794840c63ac65e48470262f7c5d4f4a0aba993bdb32856060c0fea5)
• Memory
  • DuckDB WASM support (1999837)
  • DuckDB WASM Node.js runtime support (1999837)
  • Persist DuckDB WASM data (feat: fs support with drizzle and duckdb-wasm #30) @sumimakito
  • Drizzle ORM driver for DuckDB WASM (a48624e)
  • DuckDB WASM Vector PoC (ab893fc)
  • DuckDB WASM Graph transversal (fba1029) (see also: Make DuckPGQ loadable in DuckDB WASM cwida/duckpgq-extension#170)
  • DuckDB WASM RAG
  • Short-term memory decay (read also: Efficient Points Decay for User Rankings, and Simple Feed Ranking Algorithm)
  • Long-term memory
  • Time awareness (scheduling & planning)
  • Fix Apache Arrow's (upstream) support for MONTH_DAY_NANO intervals @sumimakito
• Body
  • Talking head @sumimakito @nekomeowww (read also: https://github.com/warmshao/FasterLivePortrait)
  • Live2D
    • Motion editor @LemonNekoGH @nekomeowww
    • Motion embedding & RAG from semantic @nekomeowww (still open for other co-worker)
  • VRM
    • LipSync
• Inputs
  • Fix Discord Voice Channel STT (speech recognitioning) not working
  • Telegram video messages
  • Discord text messages
  • Perhaps design a IM layer? (Perhaps take a look at Koishi)
  • Default included Speaches (a.k.a. fast-whisper-server) for STT
  • Microsoft/Azure AI support for unspeech (feat: support Microsoft Azure AI/Speech service unspeech#15)
  • Better TTS with provider options
• Intelligence
  • 🤗 Transformers.js provider
    • Embedding (6d4b534)
    • Support loads of providers
    • MCP support (wait for feat: new package @xsai-ext/mcp xsai#84)
    • Browser use
      • browse x feat: x services #43
      • browse YouTube
• Games
  • Factorio PoC airi-factorio @LemonNekoGH
  • More games? What games wanted?
  • Improve Minecraft with newly added DuckDB WASM memory layer @luoling8192 @nekomeowww
  • Minecraft with CV capabilities
  • Minecraft mod support (make it playable, try stream with it)
  • Improve Factorio (make it playable, try stream with it)
  • Factorio mod support
• User experiences
  • New demodel component to boost speed of loading models, datasets, images from 🤗 Hugging Face and Ollama
• Community Support / Marketing
  • Issue templates, code of conduct, security, Pull Request templates, contribution guidelines
  • Stream with Airi (DevStream)
  • Building Discord server (https://discord.gg/DDetES58eK)
  • Writing docs to introduce what we have done
  • Preparing docs for new come users
  • Write DevLog

v0.5

• Memory
  • pglite RAG
• Games
  • Connect to Nintendo Switch (read also: https://zenn.dev/noov/articles/8afaf41678dfa7)

v0.6

• Memory
  • Postgres RAG