Last week I said I'd learn how to build a ChatGPT app w/ OpenAI SDK and make a demo app. I did and here it is. But... Not super impressed.
It just doesn't feel like the future to build a web component and send it back to the AI UI. It's slow and... not sure if necessary?
I feel like the future of AI UI could be a lot more interesting than showing web pages inside of AI assistants.
I'm not sure if I'll keep building more UI components for ChatGPT. Also I don't want to lock myself into their world. I like MCPs that work across all AI assistants/agents.
Also, after having spent a ton of time sending apps for approvals to Apple, Shopify, Meta, TikTok - I'm not sure if I have the patience to add OpenAI app approval process added to my life.
More interested in voice agents (https://cloe.fyi), memory (https://memkit.ai) and privacy of our personal data for AI use (https://arca.fyi) right now so I want to spend more time in those areas.
Resources:
https://openai.com/index/introducing-...
https://developers.openai.com/apps-sdk
https://alpic.ai/blog/inside-openai-s...
https://github.com/alpic-ai/apps-sdk-...
• Build Your First ChatGPT App: Fantasy Prem...
https://github.com/hollaugo/tutorials...