Case Studies

Projects we've actually built and shipped.

Three engagements spanning healthcare software, production AI, and long-term international delivery. Real systems in real use — not concepts or mockups.

Discuss Your Project
HealthcareCRMCompleted

Clinic CRM

The Challenge

Clinic administrators managed patient records, appointments, billing, and revenue across disconnected spreadsheets — causing double-bookings, delayed billing, and no real-time visibility.

Our Solution

A unified, role-based Clinic CRM covering the full patient lifecycle — registration, scheduling, invoicing, and automated follow-up — on a secure PostgreSQL data layer with JWT-based access control.

The Outcome

Clinic staff now operate from a single system of record. Scheduling conflicts are eliminated and patient follow-up is automated rather than manual.

  • Patient Registrationdigital onboarding with demographic capture, medical history, and document attachment.
  • Appointment Managementscheduling, rescheduling, and real-time availability tracking.
  • Invoice Generationautomated billing with service catalog integration and tax computation.
  • Revenue Dashboardreal-time financial analytics with daily, weekly, and monthly views.
  • Follow-up Remindersautomated patient communication via SMS/WhatsApp for post-visit care.
  • Authentication & RBACJWT-based secure login with role-based access for doctors, admins, and staff.
  • Service Catalogueconfigurable list of clinic services with pricing and category management.
AngularSpring BootJavaPostgreSQLJWT
Artificial IntelligencePropTechWhatsApp IntegrationCompleted

AI Real Estate Chatbot

London & Slough Market
The Challenge

A residential property agency was losing hours of agent time to unqualified enquiries — every lead required manual assessment before deciding if a viewing was worth scheduling.

Our Solution

A production AI chatbot grounded entirely in real, scraped property data, with a proprietary 1–10 intent-scoring model, deployed natively on WhatsApp via Twilio with persistent user memory.

The Outcome

Agents now receive pre-qualified leads instead of raw enquiries. The bot filters casual browsers from serious buyers automatically, around the clock.

Key Features Delivered
  • Persistent User Memoryremembers each user's name, phone, and preferred area; returning users are greeted by name.
  • RAG Property Search206 real listings scraped from OpenRent and Foxtons; the AI injects verified listings and never invents a property.
  • WhatsApp Integrationfull onboarding, property search, and intent scoring delivered natively via Twilio's WhatsApp Business API.
  • Professional Agent Tonesystem prompt trained on real agent-client WhatsApp conversations; strictly declines off-topic questions.
AI Intent Scoring Model
1 – 3BrowsingGeneral information only, no sales pressure.
4 – 6InterestedQualifying questions to understand requirements.
7 – 10Serious BuyerImmediate push toward a scheduled viewing.
React.jsPython / FastAPIGroq (Llama 3)PostgreSQL 18Selenium + BeautifulSoupTwilio WhatsApp API
International SupportGermany-Based ClientOngoing

International Engagement

Confidential — references available under NDA
The Challenge

An established Germany-based enterprise client needed reliable, ongoing engineering capacity without the overhead of a full local hiring process.

Our Solution

A structured, agile delivery workflow embedded into the client's own sprint cadence, with data handling aligned to European (GDPR-adjacent) requirements.

The Outcome

The engagement continues today as an ongoing retainer relationship — evidencing our ability to sustain long-term, cross-border technical partnerships.

Want references?

We're happy to walk you through any of these in detail, and to arrange client references under NDA.