Find Federico on MastodonFind John on MastodonFind Alex on MastodonFind Club MacStories on Mastodon

GPT-4.1 Talker: A Proof of Concept to Enable Stateful Conversations with GPT-4.1 in Shortcuts

SHORTCUTS CORNER

Get help and suggestions for your iOS shortcuts and productivity apps.

Shortcuts Essentials

GPT-4.1 Talker: A Proof of Concept to Enable Stateful Conversations with GPT-4.1 in Shortcuts

GPT-4.1 Talker in Siri.

I’m in the process of updating some of our internal shortcuts to use OpenAI’s API-only GPT-4.1 model instead of Claude, and in doing that, I realized that it was a good opportunity to finally learn the new Responses API. Announced by OpenAI a couple months back, the Responses API will eventually replace the old Chat Completions API (if you recall, the one I used for S-GPT) and offers more options such as tool calling and, most importantly, conversation states. One of the biggest challenges of the old API was, in fact, keeping track of the back and forth between the user and ChatGPT by having to manually track and reconstruct the entire flow of a conversation every time you wanted to ask a follow-up question. With the Responses API, OpenAI’s servers can manage the state of a conversation for you and track individual IDs assigned to each message in a conversation.

With that in mind, it seemed about time I learned the new API and start transitioning some of my shortcuts to it. Today, I want to share a proof-of-concept shortcut that you can use and learn to build your own conversational workflows with GPT-4.1 or any other model available on OpenAI. This shortcut, which I’m calling GPT-4.1 Talker, has two key characteristics: it uses the Responses API, and it relies on an advanced technique to loop on itself and enable “stateful” conversations.

This story is for Club MacStories, Club MacStories+, and Club Premier members only.

Join the Club and get access now.

Already a member? Sign in