HomeUse casesInsightsAboutעבריתLet's talk

Labs

DevShift Labs. Projects and experiments with AI.

Things we build to see how far AI gets with a hard brief. Some are finished projects you can explore, others are experiments we ran and published.

A golden 3D relief map of the Land of Israel and its surroundings under a night sky, titled The Tanakh Atlas

Featured project

The Tanakh Atlas

An interactive 3D map and timeline of the whole Tanakh. Fly from Creation to the Return to Zion, watch kingdoms and borders change, and open the verses behind every event.

  • 250events
  • 38historical maps
  • 2chronologies

Built with AI agents, which also checked every event against the verses it cites.

Open the atlas

Experiments

One prompt, two AI models

We gave Claude Fable 5 and GPT-5.6 Sol the same brief: a product page whose 3D object transforms as you scroll. We answered no questions during the run and repaired nothing after it. Here are the two pages they built.

Claude Fable 5 + Claude Code

Continuum by Fable 5

One 3D object assembles itself as you scroll, in four stages: Frame, Model, Build and Ship.

Open the page
GPT-5.6 Sol + Codex

Continuum by Sol

The same brief, built by a different model: one product system takes shape in four passes as you scroll.

Open the page

How do they compare? Build time, tokens and our verdict.

Read the benchmark

Want something like this built?

Tell us what you are trying to make. We will tell you honestly what AI can do for it, and what it cannot.

Let's talk