AI development tools like Claude Code have popularised “vibe coding”, generating working software straight from natural-language prompts. For energy executives, the appeal is obvious: build bespoke software fast, without locking into rigid vendor contracts.
Early prototypes land quickly and look impressive. The friction starts when live assets like EVs, solar, heat pumps, and smart meters have to talk to legacy billing and CRM systems, and it compounds from there. Scarce engineers drift from shipping new products to maintaining the code underneath them. In an industry that every home and business relies upon for affordable, reliable power, there is little margin for error.
Faced with such high customer and regulatory standards, energy companies often extend legacy software contracts. In doing so, they trade away innovation and take on years of slow, expensive configuration work just to make an ageing system bend to each new demand. Those systems were designed for monthly billing, daily batch cycles, and a single electricity or gas account. They break when faced with 15-minute settlement, high-frequency telemetry from multiple connected devices, and power flowing in both directions.
Buy vs. build is no longer binary
A modern, secure, energy-specific tech foundation takes years to develop, and within a customer operations system it represents about 90% of the code. It is the engine. It requires a real-time data model that reconciles what today sits in siloed billing and customer care systems. It brings global products and localised customer journeys, purpose-built for the sector. It delivers enterprise-grade security, ISO compliance, audit trails, and interoperability from day one, rather than something rebuilt release after release.
The split looks like this:
The hardest part of that engine is the intelligence inside it. Training machine-learning models across dozens of device types and billions of data points is slow, costly, and never really finished. A platform that has already done it, with those models ready to call through APIs and a newly launched MCP, turns years of data engineering into a configuration decision. That intelligence isn’t one thing either. It runs from a machine-learning core out to the experiences a customer actually touches:
The inner rings (the models, the agents, the tooling) are what you buy and bring. The outer ring is what you build together. None of the engine is what makes a customer choose you or stay.
The differentiation is the 10%. That’s the layer customers feel: dynamic tariffs, intelligent EV charging, insight-led digital experiences, and agentic workflows that act in real time on a customer’s behalf, all configured to each retailer’s own rules. Unique products, services, insights, and customer experiences are what attract customers, build loyalty, and grow customer lifetime value.
You don’t have to build that layer alone. It can be co-designed and co-built by your own developers in partnership with Kaluza’s forward-deployed engineers and specialist consulting partners. The foundation carries the accountability that matters most as AI shifts from writing code to taking actions for customers, so your people spend their time on the propositions that win and keep customers.
The choice, reframed
Try to build the whole stack, and your best engineers will spend their careers on billing reconciliation, multi-market compliance, and hardware orchestration. Try to buy the whole thing, or cling to a legacy contract, and you’re left with nothing that’s yours. Buy the engine, build the layer, and the tension dissolves. Kaluza runs the coordination layer between legacy systems, customer data, and modern AI tools, while your teams and ours build on top.
That’s what an adaptable platform is for: one foundation, countless experiences. None of it works unless the retailers on top can move as fast as their customers do. The energy transition needs everyone, and this is how retailers get to lead it.
