Nocode Lisboa #5: APIs, GPT-3 & Bubble

2023-02-02 14:56

Syllabus for tonight:

  1. Explore GPT-3 API example apps

  2. Setup Bubble's API Connector and register API call

    1. manually at first

    2. then cheating using cURL import shortcut

    3. free tool to visualize JSON objects: JSONvisual

  3. Build the mini app in Bubble

    1. parse response object to populate repeating group

    2. parameterize the API call to make it dynamic from inputs

    3. commit the results to a db table

    4. chaining API calls together... 🤯

  4. Troubleshooting tools to emulate sends/receives so you can inspect payloads:

    1. To emulate sending a request: Postman

    2. To emulate receiving a request: Pipedream

  5. Show SheetAI Gsheets ext for invoking GPT-3 from within Google Sheets.

  6. "Prompt Engineering" guide of guides massive Gsheet


When to use APIs vs. Zapier/Make connectors:

  • When Zapier or Make don't have a connector

  • When # of calls will be cost preclusive to use ^^

  • When fast response times are paramount

Downsides of calling APIs directly vs. using connectors:

  • no replay protection if endpoint is down

  • more coding and error-handling required

  • requires centralized application logic vs. "pushing logic to the edges" when interacting with multiple services.

Final thoughts

  • JSON as the "rail spike" for parallelizing app dev in teams

  • value of JSON for serializing complex data in personal app dev (choke points)

  • Amazon's mandate for externalizing all services (exposing their own internal service-oriented architecture for public consumption)

  • great tutorial that I'm building upon for this demo

https://us02web.zoom.us/j/89404986017?pwd=MjdmVGlJTUFiSHVVQlNMRlN5OFZOZz09

zoom link ^^

For SheetAI discount: