In this meeting, the key discussion revolved around an interesting technical problem. One of the State Changers wishes to use GPT-3, an AI text generation and analysis system, to handle customer service requests, capture relevant data, send it to another system, and relay the response back to the customer. The idea is to use GPT-3 for conversing with the customer, extract entities, and control the information to be sent to the specific endpoints, much like how the participant is currently using Dialogflow.
However, other participants noted that GPT-3 and Dialogflow operate on different levels of abstraction. Dialogflow, a more high-level service, utilizes a combination of rules-based systems and TensorFlow for text analysis, while GPT-3 boasts superior language generation and understanding capabilities. But Dialogflow excels in its additional affordances for features like entity extraction and date connections, which GPT-3 does not inherently offer. The consensus was that while GPT-3 might be a superior alternative to the underlying technology that Dialogflow is built upon, it is not a direct swap out for Dialogflow itself. This is due to the gap in functionality that would need to be filled. However, the meeting participants noted a potential market opportunity to apply GPT-3 in live support, SMS-based customer services and appointment booking, postulating that it could make for a powerful business if developed correctly. The services "Xano" and "HighLevel" featured in conversations around possible implementation. Key skills and technologies discussed include GPT-3, Dialogflow, Xano, HighLevel, API endpoints, entity extraction, rules-based systems, TensorFlow, CRM, and text generation models.
(Source: Office Hours 11/14 )
