What does NiceSalt do?

The thinking and the building. Positioning and messaging, the product, platform, or site itself, the content around it, and the measurement that tells you whether any of it worked.

Who do you work best with?

Founders, operators, and teams building something complicated, where credibility is part of the product. That has meant a physician's publishing platform, a harm-reduction health product, an AI media concept, and multi-tenant platforms inside much larger companies. Health, science, and consumer products are where most of the recent work sits.

How is this different from an agency?

You work directly with the person making the decisions. Fewer hand-offs, faster proof, and specialists brought in when the scope earns them rather than staffed onto it by default. Most projects can start within two weeks.

Someone referred me. What's the fastest path?

Request a free teardown with a link and one line about what's bothering you. David reviews every request and will confirm availability and timing if it is selected.

A working demo beats a deck

I had a thesis about audiobooks I couldn't get anyone to react to.

Every app in the category treats AI as a feature. A synthetic voice here, a recommendation engine there, sitting on a catalog as fixed as it was in 1995. The more interesting question is what happens when the story itself is malleable. I said that sentence out loud a few times and watched people nod politely.

So I built it instead. Bibo. You can open it on your phone right now.

In Bibo, Moby Dick isn't a file. It's five: the classic narration, a shortened cut for a commute, a modern YA rewrite, a sci-fi retelling, and a Spanish literary translation, all fully voiced. One asset, five products.

The nodding stopped. People started arguing about which version they'd actually listen to, which is a much better conversation than the one I'd been having.

Why the demo works and the deck doesn't

A deck asks people to imagine, and everyone imagines something slightly different, so a room of six produces six private versions of your idea and one polite consensus. Nobody can picture how it feels to drag a slider and hear a story compress. You have to hand them the slider.

A demo is also harder on you than a deck is. Bibo made me find out what AI narration sounds like across a whole book, which is not what it sounds like in a twenty-second sample. Some of it is worse than I'd have claimed on a slide. I'd rather learn that in week three than after someone commits real money.

What changed recently is the cost

This used to be a quarter and an agency invoice. Bibo went from concept to a polished, self-contained demo in weeks: AI-assisted build, AI narration, the translations, and more than fifty covers under human art direction. Which means the excuse for spending six weeks arguing about an abstraction is gone.

The pattern generalizes past audiobooks. Any organization sitting on a library, a publishing backlist, a course catalog, years of research, is sitting on versions it has never shipped. Other lengths, other languages, other audiences.

But I'd stop reading here and go open Bibo. That's the whole argument.