Custom Integrations

What does an interface audit collect?

Commissioning needs evidence. We record model and firmware, physical ports, protocols, authentication, data objects, command rights, rate limits and vendor support; a bench test is used when documentation leaves an important gap.

Requirements come first. The audit ends with supported, unsupported and uncertain functions. Uncertainty is priced as investigation, not disguised as a promise. The written result should name the model, interface, test evidence and next investigation needed before any uncertain function becomes a promise on site.

Who owns the bridge after handover?

Failure needs a tested path. The scope names the software or gateway, configuration backup, credentials, update policy and recovery steps; custom code also needs a repository and a defined runtime environment.

Vendor API changes are an ongoing risk. A manual alternative keeps the building usable while that risk is resolved. The fallback should remain documented and usable without the changed interface while a supported replacement, update or rollback is evaluated on site.

API, gateway or relay?

Interfaces need proof. An API can expose rich status and commands; a certified gateway maps known protocol objects; a relay offers a small but often dependable binary interface. The simplest interface that meets the need is usually easier to support.

How is a proof of concept tested?

Commissioning needs evidence. One representative device is connected, commands and feedback are logged, error states are forced, and restart behaviour is observed; production design follows only after those results are recorded. The test record names firmware and interface versions.

When should an integration be rejected?

The room changes performance. Reject it when the only path depends on screen scraping, a personal account, undocumented cloud calls or a safety bypass; clever is not the same as supportable.

How is integration effort estimated and kept supportable?

Scope sets the budget. Cost is driven by interface quality, number of commands and states, authentication, rate limits, event behavior, test environments, exception handling and future version changes. A supported manufacturer gateway is usually safer than reverse-engineering a private interface. Sometimes two separate user interfaces are the honest alternative when the systems have no stable, documented way to exchange data.

Commissioning needs evidence. Start with an interface proof using real hardware or a vendor sandbox before promising the full workflow; freeze a data map, command authority and timeout behavior, then test normal, delayed, rejected, duplicated and unavailable responses. Handover includes configuration, credentials ownership, version information, logs and a way to disable the bridge. If either side fails, commands should time out visibly and avoid repeated unsafe retries. Routine service should review API or firmware changes before applying them to production.

What does integration commissioning prove?

One successful command is insufficient. Commissioning exercises normal, delayed, denied, duplicated and unavailable responses with real supported interfaces, then records timeouts, authority, logs, disable controls and the expected state on both sides.