Skip to content
Back to projects

Case study 03 / 18 · Websites & apps

Rekndl

Status: Pre-launch testing and controlled rollout

An AI-assisted platform for activating new leads and re-engaging dormant opportunities. It brings campaign preparation, connected sender accounts, controlled outreach, conversation monitoring, and meeting booking into separate operator and client experiences.

Visit live site (opens in a new tab)

Case study

Rekndl combines a public marketing site with an administrator application, a scoped client portal and backend services. Approved lead lists move through import, campaign setup, review, controlled execution and reporting.

The problem

Coordinate everything between importing a lead list and handing a qualified meeting to a human sales team, with review points and client-specific access along the way.

What was built

  • Public enquiry and demo-request flows.
  • Spreadsheet preview and field mapping, campaign setup, connected sender accounts and client-scoped dashboards.
  • An AI Control Center, test-batch review, agent monitoring, meeting booking and campaign reporting.

Key decisions

  • Marketing, administrator and client experiences are separate applications.
  • Handing over a campaign prepares it; it does not give permission to send.
  • Sending requires readiness checks, approved batches and rollout switches that default to off.
  • Provider credentials stay on the server, and each client sees only their own data.

Where it stands

The marketing site is live. The platform is in pre-launch testing with a controlled rollout: live outreach runs only for explicitly allowlisted clients.

How it was built

Frontend

  • React
  • Vite
  • React Router
  • Tailwind CSS
  • Framer Motion
  • Lottie — Landing-page animations
  • Recharts — Campaign and client reporting
  • Tiptap — Rich-text editing in the administrator app
  • dnd-kit — Drag-and-drop in the administrator app

Backend

  • Node.js
  • NestJS — API for campaigns, sender accounts, booking and reporting
  • TypeORM
  • Swagger — API documentation

Data

  • PostgreSQL

AI

  • Ollama — Default provider in the AI model registry, running Llama 3 8B locally
  • Hugging Face Inference API — Alternative provider in the model registry
  • OpenAI API — Emergency fallback only

Integrations

  • n8n — Agent workflow orchestration and events
  • Google OAuth — Sign-in
  • Twilio — Webhook validation for legacy sender accounts

Infrastructure

  • Docker Compose

Testing

  • Vitest
  • Jest
  • Testing Library
  • Playwright