Just spent the last two days forcing Synthesia to do something it *should* be good at: a customer onboarding video with a clickable FAQ chapter menu. The result? Actually functional, which is a minor miracle given my track record with these platforms.
The process was less "intuitive AI" and more "manual configuration hell," but it works. Here's the gist:
* Built 8 separate FAQ answer clips, each with a different AI presenter (switching them up to keep it visually distinct, because let's be honest, one avatar for 10 minutes is a sleep aid).
* Used the "Scene" feature as de facto chapters. Naming them clearly is key.
* The real trick was in the hosting platform (not Synthesia's). I used a custom player that reads the scene metadata to generate the interactive chapter markers. Synthesia exports a single video with the scene data embedded, but you need third-party tools or a dev to make the interactivity pop.
Where the cracks showed:
* Any edit to a single FAQ answer after the fact meant re-rendering the *entire* long-form video, burning through credits. No modular "update clip 3 only" option.
* The lip-syncing on technical terms still wobbles. "Reciprocal procurement framework" came out as "recipro-cal pro-cure-men frame-werk" in a very unsettling, robotic way.
* Their built-in "interactive video" features are still toddler-stage. You're essentially paying for the avatar generation and smooth editing timeline, then building the actual functionality elsewhere.
It's a solid B- project. Useful for a scalable, always-available FAQ that doesn't require a human to record 8 versions. But it's not the all-in-one interactive video suite they're marketing. You're stitching together at least three different tools to get there.