Apr 1
Add Step Images and TTS Avatars
The Chef AI is an awesome addition and will keep people coming back to use the app. I'd add in some graphics for each step on the list (easy to follow images of 1 cup of milk, 1 tsp of butter, etc.). And maybe generate a Heygen avatar and use elevenlabs dialogue model for text to speach narration, that would make it go from good to great. I've had some good results with video / audio generation using Claude Code and the https://fal.ai API and the ElevenLabs api together. Just tell claude which model you want to use and specify the dialogue model for elevenlabs (sounds more human).
Pending
I haven't found a cheap way yet to control API costs; I just set small budgets. There are a couple video api's like the ltx-2-19b/distilled, that cost about $0.0025 per second... I use that for shorts. For the cooking part, you could repeat many of the images, especially those showing measurements and similar details. You could try having it generate many images initially, then use a cache that first checks if an image for that recipe already exists from similar ones. For the voice maybe try this out: https://github.com/remsky/Kokoro-FastAPI. Seems free.