Case study · Localstreet
Localstreet.
AI-enabled hyperlocal commerce connecting neighbourhood merchants to digital demand through reliable mobile products and real-time operations.
- Role
- Co-founder & CTO
- Period
- 2023 — now
- Flutter
- Node.js
- TypeScript
- PostgreSQL
- Redis
- WebSockets
- AI agents
The problem
Neighbourhood commerce is operationally rich but digitally fragmented. Inventory changes on physical shelves, merchants work across mixed levels of connectivity and software adoption, and customers still expect accurate availability and immediate order feedback.
Product and technical scope
My role spans product architecture, Flutter applications, real-time services, data models, observability, and the engineering workflows required to ship the system end to end.
- — Merchant and consumer experiences across Flutter and web
- — Live order and inventory propagation
- — Offline-aware workflows with explicit recovery
- — AI-enabled steps across organisational operations
Architecture decisions
Durable state remains authoritative while WebSocket events make changes visible quickly. Conditional transitions and idempotency protect order workflows, while merchant-facing interfaces distinguish pending intent from confirmed state.
What reliability means here
Success is not just API uptime. It includes inventory freshness, merchant acknowledgement time, order-state convergence, reconnect recovery, actionable exception queues, and an interface that remains honest when the physical world disagrees with digital state.
AI as an operating layer
AI is applied to concrete organisational steps where it can reduce repetitive work, improve decision support, and help teams move from signals to action. Consequential changes retain explicit review and accountability boundaries.