Skip to content
← ALL WORK
BUILDINGAI / SAAS · 2026

Revoria

An AI customer capture and conversion platform for small businesses. It catches missed inquiries, books appointments and runs follow-ups.

ROLE
Founder, product designer and builder
TECHNOLOGY
Next.js 16React 19TypeScriptPostgres (RLS)Tailwind CSSVitestPlaywright

PROBLEM

Small businesses lose customers in the gaps: a call goes unanswered, a web enquiry waits hours for a reply, and a lead goes cold before anyone follows up.

OPPORTUNITY

Front-desk communication is repetitive and time-sensitive. Software can respond at once, collect the details, book the slot, and hand over to a person when it matters.

CONCEPT

A dashboard for leads, conversations, appointments, automations and analytics, plus a public site with an interactive demo. The demo uses a rule-based receptionist for a fictional barber studio, so it runs without an LLM and never touches the database.

RESEARCH

Built from firsthand experience of service operations, where slow replies cost bookings every day. The demo scenario is fictional, so it does not claim real customer data.

DESIGN

A protected app with a dashboard, leads, conversations, appointments, automations and analytics. Charts are dependency-free SVG with keyboard support and table views for accessibility.

DEVELOPMENT

Next.js 16 with the App Router and React 19, server-first. Every data call goes through a single repository layer with per-tenant Postgres row-level security. Requests pass through a cookie check and an HMAC-verified session. Demo visitors each get a private, sample-data organisation that expires after six hours.

CHALLENGES

  • Multi-tenant data isolation: real accounts and demo orgs share one schema, so row-level security carries the weight.
  • Keeping the demo useful without an AI model or a live database.
  • Deciding what to simulate: follow-ups are simulated, not sent, so no real customer receives a message during testing.

ITERATIONS

  • Public site and rule-based demo
  • Lead and appointment capture, then the protected dashboard
  • Automations and analytics, with follow-ups still simulated

CURRENT STATUS

An MVP. The public site, demo, lead and appointment capture, and dashboard work. Automations are simulated. Payments, real email and SMS, and a real LLM provider are not built yet.

LESSONS

  • Speed to response is the value, so the product story has to lead with outcomes rather than AI.
  • Handoff to a human is as important as the automation itself.
  • Being explicit about what is simulated builds more trust than hiding it.

NEXT STEPS

  • Connect a production Postgres database
  • Add Stripe checkout with signed webhooks
  • Add an email provider for follow-ups, then SMS
  • Add calendar availability and a real AI provider behind the existing interface
NEXT PROJECTResto Master →