Patch

The Challenge

Patch targets growers and producers who need reliable environmental insight — not a demo that collapses control, operator tools, and upstream traffic into one brittle process. The hard part is scaling a serious edge stack: different security and release cadences for tight control loops, human-facing surfaces, and fleet backhaul; realistic cost and data-load expectations before hardware hits the field; and identity and supply-chain concerns that have to work on real networks for non-engineering users.

Our Approach

• Three explicit boundaries — control loops (local, deterministic behaviour that does not depend on the UI or cloud to stay safe), operator surface (where people configure, observe, and recover), and fleet backhaul (hubs, coordinators, identity, and telemetry upstream) — instead of a single overloaded binary.
• A Rust edge control plane in our patch stack: optimised services for offline operation and cloud orchestration, so the runtime stays small and failure modes stay understandable on gateway-class hardware.
• Deployment shapes modelled in software — standalone nodes, hub-connected sites, and hub coordinators — so we can profile behaviour without forcing every topology into one artefact.
• Benchmarking and simulating multi-device and multi-hub layouts before field rollouts, so we (and future customers in similar topologies) can see cost and data-load profiles early — not after the first busy season.
• Active testing with local users on real devices and networks: refining identity discovery through blockchain, tightening data flows, and proving the product UX with people who will actually run a grow, not a lab script.

The Outcome

Patch is in active delivery — an end-to-end Pendulum product that proves how we ship device-adjacent systems: hard boundaries at the edge, honest load and cost modelling before rollout, and field feedback driving blockchain identity, data paths, and UX. The same patterns inform client work where edge services, gateways, and TypeScript/React product UIs have to line up.

Key Takeaways

• Separating control loops, operator surfaces, and fleet backhaul beats a monolithic edge “demo” when you need to scale for real deployments.
• Simulating multi-device and multi-hub topologies early surfaces cost and data load — invaluable before producers commit hardware to the field.
• Internal product work is the fastest way to pressure-test IoT, blockchain, and UX together — so client delivery is grounded in practice, not slides.