Case study · Clean energy, telco & fintech

One app, one checkout and one monthly invoice for every household utility.

How we built Alfred’s end-to-end product: a unified basket checkout across 31 product combinations, member-facing iOS and Android apps, vendor integrations across 10+ legacy utilities, an AI-powered customer concierge and an observable cloud runtime.

Alfred Energy dashboard
Alfred Energy · CleanTech
  • Unified checkout
    31

    Valid product combinations in a single basket

  • Vendors
    10+

    Utility providers behind one seamless experience

  • Issue isolation
    < 60s

    To identify the affected customer, supplier endpoint and stack trace

  • Credit decisioning
    99.98%

    Successful first-time decisions across all 31 permutations

At a glance

Client / partner
Alfredalfred.co.uk
Industry
Clean Energy, Telco & FinTech
Services offered
  • Full-lifecycle Pod
  • Native Engineering
  • Event-Driven Cloud Architecture
  • AI Workflow Implementation
Tech stack used
  • React Native
  • Python (FastAPI)
  • AWS Bedrock
  • Go Microservices
  • Kafka
  • OpenTelemetry

Alfred is a high-growth UK consumer platform reinventing how households procure and manage essential utilities: broadband, mobile SIM, green energy and home insurance. Members run their entire home footprint through one app, one consolidated checkout and one monthly invoice.

Alfred partnered with us to build its complete end-to-end product architecture and engineering capability.

Introduction

UK households are perpetually burdened by fragmented billing, opaque tariff structures, and multiple distinct providers across broadband, energy, and telecom. Alfred set out to collapse this friction into an unbundled, pass-through membership model: zero markups on kilowatt-hours or gigabytes, funded solely through an honest monthly subscription.

Translating that simplicity into software meant solving an intricate matrix of regulatory guardrails. Energy (Ofgem), telecom (Ofcom) and insurance (FCA) each have divergent KYC rules, credit scoring thresholds and billing mechanics. More than a dozen external providers had to become one real-time customer journey, built cleanly from greenfield — against an aggressive launch deadline backed by tier-one venture investors.

Alfred needed a software delivery pod that could step in, take autonomous architectural ownership, and ship a production-grade platform without slowing velocity.

The challenges

Building Alfred wasn’t just about elegant user flows; it meant resolving deep structural knots across distributed third-party APIs and mission-critical compliance boundaries:

  1. Four regulated verticals, one seamless experience

    Energy, broadband, mobile and insurance operate under distinct regulatory codes (Ofgem, Ofcom, FCA). Each requires its own disclosures, contract timing, statutory cooling-off logic and cancellation procedures — yet must feel like one cohesive consumer transaction.

  2. Checkout and basket complexity (31 product variations)

    Consumers can buy any permutation of four products in a single basket. 31 valid combinations exist, each with different payment routing, asynchronous credit authorization and separate legal-entity merchant IDs.

  3. Cross-cutting credit scoring and eIDV logic

    A single Equifax credit inquiry has to be read differently by product: energy cannot decline service outright by law, mobile requires strict upfront handset fraud screening, and broadband requires address-level line qualification.

  4. Vendor orchestration at scale

    Over a dozen upstream utility APIs — from Seaglass energy industry flows (MHHS-native) to Openreach/CityFibre broadband and VMO2 mobile provisioning — all abstracted into zero user-facing latency.

  5. Fragmented customer state hazards

    Without a unified event bus, billing statuses, meter readings and address updates risked drifting apart across separate vendor CRM instances, exposing members to duplicate charges.

  6. Speed with rigorous governance

    A greenfield build racing toward a hard commercial launch, without sacrificing automated testing, SOC 2 compliance or GDPR auditability.

The approach

We engineered a modular, event-driven runtime that absorbs upstream complexity in the cloud, so the consumer never feels it on mobile:

Mobile-first platform architecture

Built natively in React Native with a fast Python FastAPI core and Go microservices for event settlement. The member UI is split cleanly across Alfred’s four utility pillars, backed by shared high-availability modules: membership tokens, payment vault, consent ledger and continuous telemetry.

Unified checkout with credit architecture

Instead of a separate checkout funnel per utility, we unified the cart. One atomic operation provisions direct debit mandates, runs non-invasive background credit checks, evaluates cross-vertical affordability and settles payment across GoCardless and card providers instantly.

Vendor integration layer

Every supplier sits behind resilient adapter pipelines. Seaglass handles Ofgem market switching; Fibre Cafe orchestrates broadband availability across Openreach, CityFibre and Community Fibre; and VMO2’s wholesale clearinghouse manages eSIM activation in under 4 seconds.

Canonical data model and state control plane

A unified PostgreSQL and Redis event store is the single source of truth for onboarding status, active line provisions, billing schedules and statutory notices.

Architecture blueprint

The systems behind the simplicity

How high-velocity engineering turned five distributed enterprise subsystems into one frictionless swipe.

  • Unified Checkout

    31 valid dynamic basket combinations executed atomically with zero double-charge risk.

  • Many Vendors, One App

    More than a dozen utility APIs abstracted behind clean REST/gRPC interfaces.

  • One Source of Truth

    A single real-time control plane keeping every customer record in sync.

  • Consent Ledger

    A cryptographically verifiable audit trail of member consent that every upstream vendor must honor.

  • Alfred Care (AI Agent)

    AWS Bedrock-backed assistant solving 78% of tier-1 support queries, with warm human escalation.

  • Telemetry in <60s

    Full OpenTelemetry tracing that spots supplier API slowdowns before customers submit tickets.

Consent engineered as a first-class entity

A three-part immutable consent stack governed by the canonical data model. Members’ explicit preferences drive downstream cookie tracking, third-party marketing suppression and real-time CRM updates via Customer.io webhooks.

Production-grade observability and security

Structured distributed logging, OpenTelemetry tracing and Sentry exception capture across all microservices, on one SLA: identify the affected customer, supplier endpoint and stack trace in under 60 seconds.

An AI-native delivery operating system

Product briefs were machine-readable, with business rules, acceptance criteria and edge cases clearly indexed. Bi-weekly releases, daily standups, automated PR reviews and AI-assisted linting kept velocity high.

The outcome

The platform is live in production, board-ready, and built to scale horizontally across the UK market:

  • Live waitlist & member product: waitlist acquisition engine, live utility management dashboard and fully integrated checkout shipped to the Apple App Store and Google Play on schedule.

  • Checkout complexity solved: zero basket collisions across all 31 permutations, with 99.98% successful first-time credit decisioning.

  • Sub-minute issue resolution: P99 mean time to detect supplier API drops fell from hours of manual triage to under 60 seconds.

  • Full statutory compliance: zero audit flags across FCA, Ofcom and Ofgem onboarding simulations.

  • 31

    Product combos in unified checkout

  • 10+

    Vendors behind one seamless experience

  • <60s

    Time to isolate any production issue

Technology used

The toolchain behind the Alfred member experience:

Mobile & WebReact Native, TypeScript, Tailwind CSS, Next.js 14 (member portal & acquisition web)
Backend & DataPython (FastAPI), Go (high-throughput microservices), PostgreSQL, Redis, Apache Kafka
AI & Member CareAWS Bedrock (Fred AI Concierge), Anthropic Claude 3.5, Customer.io, Amazon Connect
Credit & IdentityEquifax (eIDV, KYC verification, real-time credit decision engine)
Energy ProtocolsSeaglass Cloud, MHHS (Market-wide Half-Hourly Settlement) UK data adapters
Telecom & FibreFibre Cafe (multi-network gateway), Openreach, CityFibre, VMO2 Wholesale eSIM network
Observability & SREOpenTelemetry, Grafana Tempo, Sentry, AWS CloudWatch, Datadog APM

More success stories

All success stories
  • Travel Tech

    Travly

    Real-Time Booking Engine & AI Concierge

    Read case study
  • EdTech

    edX Platform

    Interactive Video Learning & Adaptive Analytics

    Read case study
  • Fintech

    Aura Finance

    Next-Gen Wealth Management & Bank-Grade Security

    Read case study
High-velocity partnerships

From introduction to proposal in days

We remove procurement drag. Partner with engineering pods ready to architect, integrate and ship without endless discovery overhead.

  1. 1

    Discovery call

    Our solutions architects review your technical requirements and schedule a 30-minute deep-dive within 24 hours.

  2. 2

    Expert architectural input

    Principal engineers diagnose your bottlenecks, mapping pod structure, technology trade-offs and compliance checkpoints.

  3. 3

    Fixed-sprint or pod proposal

    You receive an unbundled, clear proposal with dedicated named talent, velocity metrics and guaranteed SLAs.

Trusted by engineering leaders worldwide
  • Careem
  • edX
  • KAYAK
  • Insurify
  • World Bank

How can we help you build?

Tell us about your product roadmap, tech stack or scaling challenges.

  • Strictly confidential
  • Direct access to architects

Ready to start
working together?

Tell us what you’re building and where it’s stuck. We’ll tell you honestly whether we’re the right team for it.