Adding a User Interface to ChatGPT with BeeWare — Part 4
Using a LLM like ChatGPT to augment your app functionality is not without issues. For us, the main ones are availability, scalability, and reliability. If you are charging money for your software then your reputation is closely coupled with that of any services that you rely on.
We still haven’t gotten over Parse. Parse was a mBaaS (Mobile Backend as a service) platform, which Facebook acquired in 2013 for $85 million. Just prior to this we had decided to use Parse for one of our apps, and the purchase by FB seemed like a validation of the service. Facebook shutdown Parse in 2017 due to a change in strategy. The lesson is be careful who you trust and always have a plan B. There are other considerations when choosing a third-party LLM to integrate with.
- Complexity: Integrating an LLM into your application can be complex, requiring technical expertise to handle data exchange, API calls, and error handling.
- Limited Control: You might have limited control over the LLM’s functionality and behavior. Updates or changes made by the LLM provider could potentially break your app or require adjustments on your end.
- Pricing: Depending on the LLM provider’s pricing model, using an LLM service can add ongoing costs to your application, especially if your user base grows.
