A quick demo of how I’ve used ChatGPT and n8n to automate a workflow of a passion app Craftchat. This workflow makes use of ChatGPT as a reasoning gate to resolve if it ought to ship a textual content or picture response to the person. It does this by way of a mixture of immediate engineering, immediate injection, and chaining.
Conventional voice assistants use pure language understanding (NLU). They’re educated manually (supervised), use entity detection, and are guidelines primarily based. LLMs endure unsupervised coaching and use reasoning primarily based on the context of a dialog, very like how folks would. Whereas each approaches use AI, they make selections basically in another way.
Additional to this work my plan is to proceed exploring the boundaries of AI reasoning in additional advanced selections making situations, particularly for enterprise.
Merchandise used on this demo embrace: ChatGPT (chat completion), n8n (workflow), Secure Diffusion (picture era), GCP (internet hosting), Flutter (app) and Elevenlabs (TTS).
