Rochee - Home energy, simplified.
Australians are switching to electric vehicles fast. According to the Electric Vehicle Council, battery EV and plug-in hybrids make up 26.7% of new vehicle sales from January to July 2026, more than double their shares in the same period in 2025. Many owners already have rooftop solar through government schemes but their cars, roof and grid connection do not talk to each other.
BACKGROUND
The Problem
Homeowners with solar and EVs are sitting on more affordable energy that is not easily accessible or managed. Charging when solar output is high or when grid rate are low can cut the costs of every kilometre, but working it out means juggling between apps, electricity plans and EV car apps. Due to the complexity, users value convenience and often default to plugging it in once they arrive home when solar has dropped off and grid cost is typically at its peak time.



DESIGN CHALLENGE
Visualising Data & Flows
With the flows and the nature of this industry changing by the minute, users need to see where their energy is coming from and what it costs at a glance. Rochee turns complex data into clear, easy to digest information.
Earning Trust
Smart charging only save money if people allow and trust it to make changes. This means handling control of their systems to an algorithm. Rochee provides transparency and explains decisions to allow for users to take control back instantly.
APPROACH
Trust Ladder for Automation
Rather than switching and forcing users into a charging flow, trust is built first so users have confidence and knowledge in the product through:
Inform: Showing comparison of charge now versus later and its costs.
Suggest: “Charge at 11am to use your solar and save $4.20,” with one-tap approval.
Automate: Once users understand the savings and methods, it give them the power and option to handle them with a visible "why" and ability to easily override if need to.




INCORPORATING AI
Specs Claude Can Build From
I defined the foundations of the design system of typography, colour, spacing, shape, motion and core components through .md files. These became the rulebook and guidelines for Claude to follow when generating UI, copy writing and creating new components so outputs can be aligned with brandings and expected designs.
Stress Testing the System
Before generating any components or screens, I had Claude interview me to fill the gaps of a scalable system. Three key area the ensure the design system was suited for the end product, including type of platform, brand personality and colour roles.
Where I Stepped In
Claude produced fast generation in iteration and wireframe, however it would have gaps in the user needs. I reviewed every screen against three questions: Does it cover what out users need, does it build trust and is the hierarchy clear?
Often the generation is linear and did not touch edge cases. The first version only showed the actively charging state. In reality, the system may be unplugged, paused or dropped or fully charged. I designed additional states with validations to guides the users through the whole process and prevent possible mistakes.
Version Control
I tracked the specs and components in Git which gives every change to a version history. This allowed me to experiment different designs and roll it back cleanly when they didn't work.


VALIDATION
Competitive Analysis
Since Rochee hasn't been launch, competitive analysis was one way of learning user behaviour and expectations. I reviewed leading EV-charging and solar apps across three main areas, their core features, data visualisation and flows. I grouped these findings into what they did well but also their weaknesses that shaped my decision making.
Predictive Heatmapping
Heat mapping showed that the solar card pulled attention away from the primary CTA due to the strong colours used. This insight led to a review of the homepage and design change of specific components and restored visual hierarchy.
WCAG Audit
Due to the nature of dark mode, it opened up more accessibility challenges. I used Claude to audit the design system and consolidate the findings based on highest to lowest impact. These insights led me to prioritise design changes around contrast and passing WCAG level AA with only one component at 5.6 contract ratio at lowest.
RESULTS
Impact Before Launch
Accessibility: I resolved 5 WCAG issues, with all text now meeting AA contrast requirements in dark mode.
Hierarchy: After redesign and review of components and colours, the follow up heatmaps showed attention returning to the primary actions.
Comprehension: Users understand key metrics that was improve after copywriting and tone reviews.