Brand in. Playable out. Four stages you can see all the way through.

Stage 01

Pick the brand from a single link

Paste a homepage, an app store listing or a campaign page. The analyser reads the page — and, when it is blocked, falls back to about pages, manifests and readable mirrors — to pull the product story, tone of voice, logo and colour palette. The brand review step shows exactly what it found, and every field is editable before anything is generated. Optional assets (logo, product shots, key art) can be dragged in and cropped at this point too.

• Palette, logo and product context inferred from the source
• Brand review dialog — correct the summary, colours or logo in place
• Extra assets uploaded to your private brand library
• Brands are saved and reused, so the next unit starts in seconds

Stage 02

Ideas generated against that brand, not in general

Once the brand exists, the studio proposes playable concepts that fit it: the mechanic, the hook in the first second, the reward, and why it suits this audience. Retail gets basket and scratch mechanics, fintech gets choice and timing mechanics, games get the core loop. Pick one, edit it, or write your own line instead.

The brief is written before any code exists, and a build confirmation screen shows the brand inputs and the selected networks one last time before generation starts.

Stage 03

The studio builds it, and you keep editing it

Generation produces a single-file HTML5 playable per selected network — not one file retrofitted four ways. It appears in the workspace inside a phone frame you can actually play. Ask for changes in chat: a louder CTA, a slower timer, a different end card. Every revision is stored as its own version, so rolling back is a click.

• Live phone preview, portrait and landscape
• Chat iteration with full version history
• Per-network variants generated in the same run

Compliance QA runs on every build