Webinar recap: building revenue-critical IoT on APIs

When your entire business model depends on IoT data flowing reliably from hundreds of remote devices, connectivity is mission-critical infrastructure. Yet many IoT deployments treat connectivity as an afterthought, only to discover that unreliable connections, manual troubleshooting, and poor provider choices create operational nightmares that prevent scaling.

In a recent webinar, Endre Ulberg, a software engineer at Sunday Power, Norway's largest Solar as a Service provider, shared how his team built a production-grade IoT monitoring platform that keeps over 100 rooftop solar installations online and generating revenue. Joined by Parker LeBlanc, a software engineer at Hologram, the conversation revealed the real-world architecture decisions, hard-won lessons, and connectivity strategies that enable Sunday Power to scale their fleet without drowning their team in manual device management.

Whether you're designing your first IoT deployment or scaling from dozens to hundreds of devices, this recap delivers concrete insights on choosing the right connectivity model, automating fleet management, and building systems that support growth rather than constrain it.

Meet the experts

Endre Ulberg is a software engineer at Sunday Power, where he works on the IoT platform that monitors solar installations. With responsibility for ensuring that critical production and billing data flows reliably from every rooftop system, Endre brings a customer perspective grounded in real production challenges and the operational realities of running revenue-dependent IoT infrastructure.

Parker LeBlanc is a software engineer at Hologram, building tools that IoT developers use to manage their connectivity. Parker's expertise spans cellular connectivity architecture and the technical patterns that enable teams to manage device fleets with minimal friction.

Together, they provided a rare combination of customer experience and platform insight, revealing what actually works when connectivity underpins your entire business model.

Connectivity is infrastructure, not an afterthought

For many IoT companies, the temptation to treat connectivity as a commodity or rely on existing customer networks is strong—until reality sets in. Sunday Power learned this lesson early in their journey.

As a Solar as a Service provider, Sunday Power's business model depends entirely on accurate production monitoring and billing data flowing from every installation. Without that telemetry, they can't invoice customers, verify system performance, or deliver on their service promise. As Endre explained:

"Without that data, we are unable to basically get any money at all. This is our whole business model in many ways."

Initially, the team explored using customer building networks, relying on existing internet connections at installation sites. This approach seemed cost-effective and straightforward, but it quickly revealed fundamental problems. Customers controlled firewalls, changed IT providers without notice, and often had facilities where connectivity was unpredictable. The result was constant troubleshooting, manual interventions, and devices going offline at the worst possible times.

As Endre put it: "Believing that you can piggyback on someone else's network... has turned out to be false."

This realization led Sunday Power to standardize on cellular connectivity, treating it as dedicated infrastructure they could control, monitor, and depend on. By owning the connectivity layer rather than relying on customer networks, they eliminated an entire class of support issues and gained the reliability their business model required.

The lesson:

If your IoT data drives revenue, compliance, or critical operations, invest in dedicated cellular connectivity from day one. The apparent savings of using existing networks evaporate when you factor in the operational overhead, customer friction, and business risk of unreliable connections.

Simple APIs accelerate integration and reduce friction

For companies whose core business isn't connectivity, the complexity of working directly with carriers can become a significant drag on development velocity. Sunday Power needed to integrate connectivity management into their platform quickly, without becoming experts in carrier protocols or navigating fragmented APIs across multiple providers.

Endre emphasized what mattered most in choosing a connectivity partner:

"One of the most important things when picking a connectivity provider is thinking about how that is a part of your business. Solar is our main business. Connectivity is a means to an end. We absolutely need it. We're very dependent on it, but it's not core business to us."

For that reason, Sunday Power prioritized providers with straightforward integration paths. As Endre explained, they needed "a stable API... a regular REST API, no special syntaxes, no special communication patterns, nothing that is kind of no rough edges, just smooth sailing."

This approach translated to rapid integration. Endre's team could query device status and—critically—send SMS commands to devices using simple HTTP requests. No specialized SDKs, no carrier-specific knowledge, no friction.

The commercial model mattered just as much as the technical interface. Endre noted the importance of finding "deals that don't punish you too severely if it turns out that you use more data than you anticipated in the earlier phases of building your IoT product." This allowed Sunday Power to design their system for the throughput they actually needed rather than artificially constraining data usage to avoid surprise bills.

The lesson:

Choose connectivity partners whose APIs and commercial terms align with your team's capabilities and business model. Simple, standards-based integration and predictable pricing accelerate development and remove barriers to scaling.

Automated remediation transforms fleet management at scale

One of the most valuable insights from the webinar was Sunday Power's approach to automated device recovery. When you're managing dozens of devices, manual troubleshooting is annoying but manageable. When you're managing hundreds, it becomes impossible.

Sunday Power implemented a heartbeat-based monitoring system that tracks when each device last checked in. Devices report their status once per minute, and the platform surfaces any device that hasn't communicated within a defined threshold—typically around 300 seconds. This simple dashboard gives the operations team instant visibility into fleet health.

But visibility alone isn't enough. You need automated remediation. Using Hologram's API and SMS capabilities, Sunday Power built a workflow that automatically sends authenticated reboot commands to unresponsive devices. The system uses an exponential backoff schedule, attempting recovery at increasing intervals rather than hammering devices with constant commands.

This automation dramatically improved uptime. Instead of operations staff manually identifying stale devices, logging into systems, and triggering resets, the platform handles recovery autonomously. The team only gets involved when automated remediation fails, which is rare.

Endre was candid about how they arrived at this solution. The team had previously used a different connectivity provider and encountered significant issues:

"We had quite a lot of painful moments before we got to maybe 40 devices."

Those early struggles led them to build automation. As Endre recalled, "We actually had a manual routine of restarting our devices once a day" before implementing the automated system.

Now, for their stable fleet, "it's almost never down more than five minutes, and that is an acceptable range for us."

The lesson:

Build automated monitoring and remediation into your platform from the start. Use heartbeat tracking, threshold alerts, and API-driven recovery workflows to keep devices online without manual intervention. The operational leverage compounds as your fleet grows.

Design for throughput headroom and future flexibility

A common mistake in IoT deployments is optimizing for minimum viable bandwidth. Teams calculate the absolute minimum data requirements, select the cheapest connectivity option that meets that threshold, and move forward, only to discover later that they've constrained their platform's evolution.

Sunday Power took the opposite approach. Rather than minimizing data usage, they designed for throughput headroom. This meant choosing cellular modules and connectivity plans that could handle significantly more data than their current minimum requirements.

Why? Because requirements change. As Sunday Power's platform matured, they could add new telemetry, implement more frequent reporting intervals, and explore new capabilities without re-architecting connectivity or replacing hardware in the field.

As Endre explained:

"We need quite high data throughput. We have kind of decided that we would rather invest in slightly more expensive equipment and slightly more expensive site installations if that allowed us to get higher throughput and in some sort of sense, future proofing of the platform."

The choice between connectivity technologies also mattered. Sunday Power tested LTE-M for better indoor penetration in buildings with multiple layers of concrete, but found the throughput too low. Devices started throttling, backing up data, and production information began arriving with delays—sometimes a minute or two after it happened instead of within seconds. They validated that standard 4G provided the throughput and reliability they needed in their actual deployment environments, particularly given Norway's strong 4G coverage, including mountainous regions.

By combining 4G with batching and compression, Sunday Power achieved the right balance:

"Right now, we don't use a lot, but we have a lot of headroom, and that headroom allows us to kind of fearlessly just build whatever we want and see how it turns out."

The lesson:

Design your connectivity architecture for future needs, not just current minimums. The incremental cost of higher throughput is usually trivial compared to the cost and complexity of retrofitting connectivity later. Validate technology choices in your actual deployment environments before committing.

Security and operational maturity must scale with your fleet

One of the most honest moments in the webinar came when Endre discussed security for their SMS-based control channel. Currently, Sunday Power uses SMS commands with admin passwords to remotely manage devices—a pragmatic approach that works at their current scale but needs to evolve.

Endre acknowledged that as the fleet grows, this model creates risk. A single compromised password could affect multiple devices, and rotating credentials manually across hundreds of installations isn't sustainable. The team is planning to implement per-device credentials and automated secret rotation.

As Endre explained: "Right now we're in the middle of changing to a new hardware device that has a built-in modem and works in a quite different way from the current setup. And there we are going to have unique passwords per device and be a little bit more good at rotating our secrets quite frequently, and maybe also finding other ways to make the SMS control plane that we have more secure."

This candid discussion highlighted an important principle: operational maturity is a journey, not a destination. The systems and practices that enable you to launch and validate your product aren't necessarily the ones that will carry you to enterprise scale. Building in the flexibility to evolve—and choosing partners who support that evolution—is critical.

The lesson:

Be honest about where your security and operational practices need to mature as you scale. Plan the transition to per-device credentials, automated secret management, and more robust control planes before they become urgent problems. Choose connectivity partners who provide the tools and flexibility to support that evolution.

Actionable takeaways

Ready to apply these lessons to your own IoT deployment? Here are five concrete steps you can take:

Implement heartbeat monitoring

Build a system that tracks when each device last communicated, surfaces stale devices in a dashboard, and triggers alerts when thresholds are exceeded. This gives you real-time fleet visibility and catches connectivity issues before they impact your business.

Automate device recovery workflows

Use your connectivity provider's API and SMS capabilities to send authenticated reboot commands or other remediation actions automatically. Implement exponential backoff to avoid overwhelming devices while maximizing recovery success rates.

Evaluate connectivity partners on API quality and commercial terms

Prioritize providers with straightforward REST APIs, clear documentation, and transparent pricing. Avoid partners whose complexity or cost structure will constrain your development velocity or penalize you for growth.

Design for throughput headroom

Choose cellular modules and connectivity plans that provide significantly more bandwidth than your current minimum requirements. This creates strategic flexibility to add features, increase reporting frequency, or implement new capabilities without re-architecting your platform.

Plan your security evolution

Map out how you'll transition from simple authentication models to per-device credentials, automated secret rotation, and more secure control planes as your fleet scales. Start implementing these practices before they become urgent operational problems.

Building IoT infrastructure that scales

Sunday Power's journey from initial deployment to production-grade IoT platform illustrates a fundamental truth: treating connectivity as mission-critical infrastructure from day one rather than an afterthought creates the foundation for reliable, scalable operations.

By choosing cellular connectivity they could control, partnering with a provider that offered simple APIs and predictable commercial terms, implementing automated monitoring and remediation, designing for throughput headroom, and planning for operational maturity, Sunday Power built a platform that supports their business model rather than constrains it.

The result? A fleet of over 100 solar installations generating reliable data, with their stable fleet "almost never down more than five minutes", minimal operational overhead and the flexibility to evolve their platform as their business grows.

Whether you're launching your first IoT deployment or scaling from dozens to hundreds of devices, these lessons provide a roadmap for building connectivity infrastructure that just works—so you can focus on your core product instead of troubleshooting modems.