innochat updates

• New agent models: We are expanding the platform with even more powerful and versatile models. You can now choose from a wider range of options, including: gPT-O3, GPT-O3 Pro, GPT-03-mini, GPT-O4 mini, GPT-4.1, Gemini 2.5 Pro Preview, Gemini 2.5 Flash Preview Thinking, Gemini 2.5 Flash Preview, Claude-4.0-Sonnet and Claude 4.0 Opus.

• Revised citations and sources: We've improved the presentation of citations and sources to make things easier for users and fix layout issues.
• Variables for initial messages in iframe: You can now send variables to the iframe for initial messages (just like regular script-based widgets). This means that you can pass information to the chatbot before it is rendered. This opens up applications such as sending information about the user and the page they are currently viewing to your AI even before they start chatting.
• Upgrading the data collection webhook: The data collection webhook now includes a toggle option that sends all existing variables when triggered (instead of just the selected ones).

• Fixed an issue that prevented users from seeing all debug messages in Firefox.

We're introducing a new, convenient way to share your chatbots: the chatbot library. No complex integrations. No cumbersome setup. Just add your bots to a single library and invite others to use them the same way you use ChatGPT:

• 📚 Build your own library: Add as many chatbots as you'd like. The chatbots in the library behave just like ChatGPT. Users can track older sessions and revisit them as needed.
• 👥 Invite others: Share your library with a single link. Create accounts for your users who can sign in to your library without having to register with GPT-Trainer.
• 🔄 Easy access: Users can select from your list of chatbots without any configuration. You don't need to get in touch with the main GPT Trainer user interface.
• 🧵 Manage threads: Keep conversations organized and manage ongoing threads with ease. Administrators can see all sessions, while users can only see their own sessions.

• Collected data tab: Get a handy overview of the data collected by your chatbot with our new Collected Data tab. This special page provides a central view of all structured data stored in user sessions.

• Redesigning the chatbot: The appearance of our chatbots has been redesigned! This redesign includes updated styles and customization options. Now you can choose a primary color that matches your brand, discover various options for embedded widget buttons, and unlock new ways to present your AI assistant.

• New agent models: We're excited to share a powerful suite of AI models for your agents! Now available: DeepSeek-R1, DeepSeek-v3, Gemini 1.5 Pro,Gemini 2.0 Flash, Claude 3.5 Haiku, Claude 3.7 Sonnet, GPT o1 and GPT O3 mini. These models are available in various versions and support context windows ranging from 2k to an impressive 128k. This gives your agents the flexibility to handle everything from short chats to deep, context-rich conversations.

We have also improved several existing models by increasing their scope of context.

Thanks to these additions and improvements, your AI agents are now more versatile and powerful than ever before. From quick answers in customer service to detailed conversation analyses — the new models and larger context windows give you a wide range of options for adjustments and efficiency improvements.

• Variables for first messages: Customize your chatbot's initial messages with dynamic variables. Now you can attach user-specific data to customize the chatbot's initial interaction based on various input parameters. “Hi {{{user_name}}}” instead of “Hello! ”`.
• Low balance notifications: Never again unexpectedly without news credits! You can now enable low balance notifications in your account settings so that GPT Trainer notifies you when your credit is running low (applies if you haven't set up an AI API key yet).
• Adjusting the spacing between widget buttons: Adjust the exact position of your chatbot widget! You can now flexibly adjust the position of the speech bubbles to ensure optimal placement of the widget on your website.

• Additional details in debug messages: Get deeper insights into your chatbot's decision-making process. We've added RAG context, Google search results, and error logs to the debug messages so you can better understand the data sources behind the responses and troubleshoot issues more efficiently.

• API session creation: Fixed an issue where the session creation API incorrectly required a mandatory JSON payload, which caused disruptions for some users. This has now been resolved.

• ‍UTM tracking: You can now enable UTM tracking to track and analyze the performance of your campaigns. Track user behavior across marketing channels and campaigns to better understand where your traffic is coming from and make data-driven decisions.

• SMS integration: Connect your phone number to your chatbot and receive messages instantly. With this new SMS integration (in collaboration with Ytel), you can configure GPT trainer AI agents to automatically respond to incoming SMS messages.

• Exporting tracking data: We've made exporting your tracking data easy! You can now export a detailed list of all your tracking information in one convenient file.