Many fascinating things can be done by using the normal chat interfaces like ChatGPT, but eventually you may want to offer a generative AI powered experience to others, or create automated workflows within your organization that combine LLMs with other systems.
The breadth and depth of these methods is vast. We offer the two simple examples below to demonstrate the basic concepts behind generative AI app development and workflow automation.
Making a multi-modal app with Bubble:
Bubble is a no-code app development service. We will use it to practice connecting APIs to a simple frontend to create a multi-modal generative AI app…in other words, an app where a user can put in one input, and get an output in more than one medium — in this case, text and imagery.
The first step → Creating an API key for text (OpenAI) and image (stable diffusion) generators:
Developing the rest of the app
Making the app:
Adding images to the app:
Using the lastest instruct models in your app:
Building automation workflows with Zapier:
This is an instructional on how to use a workflow automation tool Zapier in order to connect AI with other software. Zapier allows different apps to interact with each other. Recently, this enabled Zapier to connect ChatGPT with apps like Medium and LinkedIn, allowing generated text to be automatically posted in these social media sites.
The following steps show how to connect:
- Zapier→ Email. Email content sent to Zapier triggers a chain of emails.
- Email → LinkedIn. Content sent in an email is automatically posted in LinkedIn
- Email → ChatGPT. Content sent in an email is automatically summarized into a Medium blog post by a Large Language Model (LLM) ChatGPT.
- ChatGPT → Medium. Content generated by ChatGPT is automatically posted on Medium.