Selora AI v0.6.0: Google Gemini, Proactive Chat, and Smarter Suggestions

Selora AI v0.6.0 is out. This release focuses on making the AI smarter and more conversational — it can now suggest actions in real time, act on what you ask, and run on Google Gemini alongside existing providers.

Google Gemini LLM Provider

Selora AI now supports Google Gemini as an LLM backend. If you already have a Google AI API key, you can plug it in and start using Gemini models immediately.

This is possible thanks to a new LLM provider abstraction layer we built under the hood — making it straightforward for us to add more providers in the future. Configure your provider in the Selora AI settings panel, and API keys are now validated on save so you’ll know right away if something is wrong.

Proactive Suggestions in Chat

Selora AI already watches your home and proposes automations. Starting with v0.6.0, those suggestions now appear directly in the chat interface — no need to navigate to a separate panel.

When Selora AI notices a pattern worth automating, it surfaces the suggestion inline so you can review, accept, or dismiss it without leaving the conversation. Suggestions are also deduplicated by content hash, so you won’t see the same idea twice.

Act After State Queries

Ask Selora AI “is the garage door open?” and it will tell you — but now it also offers to act on the answer. If the garage door is open, it will ask if you’d like to close it. If the living room lights are off, it will offer to turn them on.

This makes the chat feel more like a conversation and less like a read-only dashboard. Queries naturally lead to actions, and Selora AI bridges the gap.

Hardened Entity Validation

Suggestions are now validated against real entities before being presented. If Selora AI proposes an automation that references an entity that doesn’t exist, has been disabled, or can’t perform the suggested action (like trying to turn off a binary sensor), the suggestion is filtered out before you ever see it.

This means fewer broken automations and more confidence that what Selora AI proposes will actually work when you accept it.

Bug Fixes

This release includes a significant batch of reliability improvements:

• Binary sensors are no longer suggested as automation actions ( #91 )
• Disabled entities are filtered out and same-device correlations are skipped ( #93 )
• Duplicate suggestions are eliminated via content hash ( #46 )
• Conversation context is retained across chat messages ( #89 )
• Device commands execute correctly from streaming chat ( #90 )
• Security hardening and memory leak fixes across the integration

See the full release notes for the complete changelog.

What’s Next

Selora AI Cloud — our own hosted LLM backend that removes the need to bring your own API key — was initially planned for this release but has been moved to v0.8.0 to give us more time to get it right. In the meantime, we’re continuing work on delayed actions via chat and scene creation .

Update via HACS, or if you already have Selora AI installed, the update will appear in your Home Assistant settings.