What’s worth building before the models are ready?

Read [Every’s Vibe Check of Astra for indie filmmakers](https://every.to/vibe-check/vibe-check-is-astra-a-breakthrough-for-indie-filmmakers). Explain what became possible for Afterimage and what still failed or required human direction. Use that example to help me think about building ahead of model capabilities. Based on what you know about my work, suggest one project, workflow, or creative idea worth preparing for even if today’s models cannot yet support it reliably. Identify the specific limitation holding it back, what I could test or build now, and what evidence would justify investing further. Separate demonstrated capabilities from assumptions about future progress; don’t assume the next release will solve the problem. If nothing seems worth that bet, say so. If you need context, ask one question.