Almost everyone seems to have integrated ChatGPT into their platforms, but it’s often just another way to generate text.

I believe the real benefit from AI API’s will come from integrating it into the user flow allowing for automated, contextually aware actions.

A user could ask the AI to do something outside of just editing content, performing manual actions automatically.

Expanding app capability

An example of this is the WordPress page-builder Elementor. On their site they promote the ability to generate text. However, if they wanted to expand their offering, they could integrate AI to perform tasks in WordPress/Elementor platform itself:

How many pages do I have?

Create a new Nav template and connect a new menu to it.

Do I have any elements that are using the custom CSS box?

Roll back all changes to last Friday.

Kind of like an extra-capable Siri (our smart assistants will also benefit from this enhanced contextual awareness).

How can I do it?

To integrate this idea into your own SAAS app, you could follow this process.

Connect your app to the ChatGPT AI. OpenAI has a great repo showing you how to do this.

Use this starter prompt:

You are a robot for my app and can only reply with certain commands. No descriptions or external information, just the commands. Based on my input you have to work out which command to return.

Only reply with these commands:

WAIT(1s)

SELECT COMPONENT(name)

CHANGE BREAKPOINT(xl)

TOGGLE_DARKMODE

INFO(msg)

Now you can build callback functions to handle each command. In this example, it’s for a UI browsing app.

Example reponses could be:

User:
Show me all the components in the Nav folder in light-mode then dark-mode.

Response:
SELECT COMPONENT(nav1.html)
WAIT(2s)
TOGGLE_DARKMODE
WAIT(2s)
SELECT COMPONENT(nav2.html)
WAIT(2s)
TOGGLE_DARKMODE

---

User:
What components have custom javascript?

Response:
INFO("The nav1.html and footer2.hmtl have custom javascript")

---

User:
How does this look across all breakpoints?

Response:
CHANGE BREAKPOINT(xl)
WAIT(2s)
CHANGE BREAKPOINT(lg)
WAIT(2s)
CHANGE BREAKPOINT(md)
WAIT(2s)
CHANGE BREAKPOINT(sm)
WAIT(2s)

Conclusion

LLM’s have the benefit of interpreting human questions and users will start to expect all their apps to intelligently identify and perform requested actions.

Developers can start to use AI now to eliminate pain-points and remove repetitive and manual tasks, making our software faster and more intuitive.

I’m excited to see how companies improve their products and services to make technology even more seamless and hassle-free.

Further Reading