System Updates - July 22

User Interface

• Able to select model priority from a drag and drop list in Bot Settings > General
• Test Bot page has changed to allow you to test your bot with OpenAI vs. Anthropic
• Toggle to turn off the default ON feature of always aware AI (discussed below)
• Changing between sub-accounts would sometimes show the incorrect bot in the dropdown compared to bot that was displayed in bot settings. This has been resolved.
• Removed ability to toggle allow fallback from User Settings page. This is now handled by modifying the model priority order within Bot Settings > General

AI Performance

• 🔥 [ New Feature ] Ability to choose bot LLM provider priority. The options include Anthropic (credits), OpenAI (credits), Anthropic (your key), OpenAI (your key). You can choose the order of these and even make it so credits aren't an option so your credits never get used. In this case, failed keys result in no AI response.
• 🔥 [ New Feature ] Always Aware AI which means the bot will now be aware of all recent messages to and from the contact (not just messages involving CloseBot). This means you can turn the bot off using ai off tag, have a conversation, turn the bot back on and the bot will be aware of what was said and update fields. This also means there's no longer a need for Override Messages
• More reliable and quicker auto-off on manually sent replies. Previously typo messages and broken up messages from the bot were falsely triggering the auto-off function.
• Updated Anthropic usage to use Sonnet 3.5 from previously used Opus