Back to projects

Help food businesses get online

Kilat.store

Help cafes, restaurants, and local food brands create a website, display their menu, and receive inquiries through WhatsApp.

Outcome

Business owners can describe their business, edit a prepared website, and publish a shareable link without learning to code.

My role

Full Stack Engineer & AI Integrator

How I helped

  • Connected website generation, visual editing, and publishing.
  • Built a process that prepares pages without making owners wait on a frozen screen.
  • Set up delivery designed to keep website hosting inexpensive.

Product

The workflow in context

Captured from kilat.store in September 2026.

Live Kilat.store homepage showing its F&B website generator.
Public website screenshot

The solution

Focus on what a food business needs first: a menu, business details, and a clear way to order. If AI is unavailable, provide an editable starter instead of a blank page.

What changed

A free first release is publicly available with one website, visual editing, a kill.lat address, and WhatsApp contact. Custom domains and paid upgrades are planned, not included in the current offer.

Lessons learned

A focused service that gets a business online is more useful than a long list of features owners have to figure out.

Behind the solution

Indonesian F&B merchants often rely on social media alone, while generic website builders make menu setup, mobile ordering, and publishing harder than necessary.

A monorepo combines the marketing site, Next.js editor powered by Puck, and a Rails API with background jobs for AI generation. Published storefronts use the same canonical block contracts shown in the editor and receive a shareable kill.lat subdomain.

Constraints

  • Must run smoothly on low-end mobile devices.
  • Generation must still produce an editable starter when AI is slow or unavailable.
  • The free MVP must keep publishing and ordering simple without paid-plan complexity.

Technical decisions

  • Used one canonical block registry across generation, editing, and rendering.
  • Moved AI generation to Rails background jobs to avoid request timeouts.
  • Added a safe fallback F&B template instead of leaving merchants with a blank screen.

Key features

  • AI drafts grounded in business type, menu, location, audience, and visual style.
  • Visual editing for copy, images, colors, and responsive content blocks.
  • Menu, reservation, ordering, QR menu, and WhatsApp action blocks.
  • Immediate publishing to a Kilat subdomain.

AI system implementation

Rails background jobs draft sites from business type, menu, location, audience, and style. Generated content uses the same canonical block registry as the Puck editor and published renderer, keeping drafts editable rather than treating model text as a finished website. A deterministic food-business template remains available when generation is slow or unavailable.

Could a similar approach help your business?

Tell me where work gets stuck. We can identify a practical first improvement and what success should look like.

Discuss your challenge