Respond AI: Support for OpenAI Assistant API
I
Isaac Meneses
Business Problem:
The current Respond AI model presents limitations in flexibility, including restricted customization options for AI personas and difficulty understanding varied customer inputs (e.g., synonyms or informal phrasing). This results in less effective conversations and reduced personalization.
Desired Outcome:
Integrate OpenAI’s Assistant API to enable businesses to deploy highly customized and context-aware AI Agents, offering improved understanding, better brand alignment, and a more tailored customer experience.
Use Case:
- Allow users to bring their custom GPTs built using OpenAI’s Assistant API into Respond.io.
- Improve natural language comprehension (e.g., recognizing synonyms like "auto" for "car").
- Customize AI persona behavior and memory to align with business goals, tone, and processes.
- Provide more flexible control over AI logic and responses compared to the current built-in model.
Amalia Putrieka
Hi Isaac Meneses thank you for your input!
To better understand your needs, could you share some specific use cases or goals you’d like to achieve with the improved versions of ChatGPT and custom GPTs? Your insights will help us explore the best possible solutions.
Looking forward to your response!
I
Isaac Meneses
Amalia Putrieka There are several issues with the current model. Additionally, the space for creating a custom AI persona is very limited. I've also noticed that a default prompt is being sent by Respond.io to verify information with the customer, which is not only tedious but also ineffective and repetitive. The AI agent struggles to understand synonyms. For example, if the function is to ask what type of insurance the prospect wants and there are three options: car, home, and business, but the customer responds with "auto" instead of "car," the agent fails to understand and will check twice to confirm if the customer meant "car." Furthermore, the AI agent frequently fails to extract information from the knowledge base.
I'd like to be able to create a customized GPT (using chatgpt) that can be accessed via the ChatGPT API, connect it with Respond.io, and create a tailored conversational experience.