AI + manual control in Trickle

Hey everyone, Bob here from Trickle. We’ve been building an AI-native platform where anyone can create apps, websites, and forms just by describing them in natural language.

But as magical as prompt-based generation is, we kept hearing the same thing from users:

“I don’t want to rewrite a whole prompt just to change the font size.”

So we built Direct Edit, a visual editing mode that lets you click on any element (text, image, layout block) and adjust it directly. No tokens spent. No regeneration needed.

Why we made this

AI gives you speed. But when it comes to polishing spacing, font tweaks, alignment, prompts become clunky.

We realized: instead of forcing users to prompt even for small things, why not just… let them click and adjust?

Now with Direct Edit, you can:

- Select any element with a single click

- Edit font, spacing, size, colors, and layout visually

- Save your changes instantly without touching prompts or using up tokens

- Stay in flow when refining pages

It’s like adding a mini Figma inside Trickle — but only when you need it.

Who it’s for:

- Creators who use AI to build, but still want control

- Designers tired of regenerating the whole page for small tweaks

- Makers looking to move fast but still care about polish

This feature just went live today. Would love to hear how other AI builders here are approaching the balance between AI generation and manual precision.

Are you building for prompt-first, or are you also finding the need for direct manipulation? How are you handling editability?

Try it out here: https://www.trickle.so/

Happy to share more behind-the-scenes if anyone’s curious how we designed the UX flow.

Let me know what you think, and if this solves a pain point you’ve also run into while building with AI.