Copilot or Autopilot?

You might be asking the wrong question.

Yesterday at Canva’s AI Vision in Sydney, someone asked:
“If you had to ship in 90 days, would you build a Copilot for today’s tasks, or an Autopilot for the ones users wish they had time for?”

The binary trap
We often get sold this choice: fast assistance (copilot) vs bold autonomy (autopilot).
Here’s the thing: I don't think that most real AI products have to choose.

The way I understand it:
- copilot solution = I help you
- autopilot solution = I do it for you
- hybrid solution = I take over when it makes sense, and step back when you need to decide

In practice:
V1: Copilot on one critical friction point (the one that actually hurts)
V2: Progressive Autopilot once trust and data are there
V3: Smart hybrid that knows when to step back or take the wheel

My 3 questions before deciding as former AI&UX team lead:
What user behavior do I want to reinforce vs automate?
Do I have guardrails that make errors recoverable?
Does my ROI come from time saved or quality unlocked?

I’m curious in your products, have you moved from Copilot → Autopilot?