Hardware is where constraints become visible.
Over eight years as the sole designer at Copeland, I led UX across five thermostat products, each with its own constraint profile, audience, and failure modes.
Some products used fixed-segment displays. Some used touchscreens. Some had three buttons and almost no room for instruction. Hospitality thermostats had to be understood from across a hotel room by guests who may not speak the language on the label.
Every surface has to work on its own terms while still feeling like it belongs to the same product family.
That is not a visual style problem. It is a decision-making problem. Every call either reinforces the system or quietly undermines it.
Scope
- Sensi Touch 2
- Sensi EIM
- Sensi Lite
- Verdant VX4
- Verdant Line Voltage
Patents
US 12,608,066
Low Power Detection
Awarded · 2026
Patent pending
Segment display character set
Photo 1 of 5
A touchscreen is not a fixed-segment display.
Sensi Touch 1 was designed to mirror the Sensi Smart, the Wi-Fi-connected fixed-segment thermostat sitting next to it in the product line.
That was a reasonable instinct for a first-generation touchscreen product. But a touchscreen is not a fixed-segment display with more pixels. It is a different surface with a different contract.
Touch 2 was the opportunity to correct that.
Touch 1's menu structure was solid, and I was not going to change navigation users already understood. But the visual language, interaction model, and interface architecture all needed to change so the product could use the capabilities it already had.
Before design work began, I analyzed Touch 1's strengths and weaknesses, mapped the competitive landscape, and interviewed utility partners, contractors, and homeowners. Each group had different failure modes to surface. The breadth was not research for its own sake. It was the only way to know which problems were real.
One button, one job.
The most consequential change was not visual.
Every screen has a primary action. In Touch 1, that action was buried or inconsistent depending on where the user was. I introduced a persistent contextual action button: whatever the most important next step was, always in the same place, always relevant to the current screen.
Users learn the pattern once, and it works everywhere.
That kind of consistency is invisible when it holds and expensive when it does not.
The architecture had to outlive the first release.
The less visible decision was building the interface architecture to be extensible.
Remote sensors and the Equipment Interface Module were not in scope at the time. I still built the UI framework so those features could be added later without breaking what existed.
They were added later, and they fit.
That is what happens when architecture is treated as a design decision rather than an engineering detail.
Physical and digital decisions were connected from the start.
I was involved from the industrial design evaluation phase, running testing and feeding back on how different form-factor concepts affected the UI.
On a touchscreen product, physical and digital decisions interact at every level. Thickness affects reach targets. Bezel width affects perceived screen space. Screen size changes what interaction density feels acceptable.
I was the connection between those conversations.
Touch 2 shipped at 0.77 inches, the thinnest smart thermostat in the category at launch. The interaction model later informed consulting work with Trane's internal design team on their premium touchscreen thermostat, which confirmed the work held up outside our own product line.
The design problem was not the hardware. It was the trip.
The Equipment Interface Module bridges thermostats to HVAC equipment in installations where standard wiring falls short.
In homes with four wires or fewer, heat pump installations may require six. Where competitors shipped three separate modules to cover the same range of use cases, the Sensi EIM handled all of them with one.
The design problem was not the hardware. It was configuration.
The obvious approach would have been configuring the EIM at the unit. But in a typical installation, that means traveling between the thermostat and the equipment, often through tight mechanical spaces or up and down stairs.
Each trip is time a contractor is not billing for.
Move configuration to where the contractor already is.
I moved configuration to the thermostat.
Through a pairing and configuration flow built into the thermostat interface, the contractor sets equipment location and type once, without leaving. The EIM receives that information and configures itself.
A workflow that used to require back-and-forth movement between two physical locations happens in one place.
That is the kind of design work that does not look dramatic on a screen. The value is in removing movement, uncertainty, and lost time from the job.
Contractors noticed the part that mattered.
The field response made the value clear.
In training sessions following launch, two separate distributor groups independently called out the pairing process as what they noticed most: configuring from one location without physically moving between the thermostat and equipment.
The launch webinar drew more than 180 contractors.
When the people the product was built for call out the workflow unprompted, that is validation.
With contractors, silence is also data. They are vocal when something does not work. The pairing flow generated no iteration requests after launch.
It just worked.
Thirty-two segments, three buttons, no room to hide.
Sensi Lite was the most constrained product I have worked on.
Thirty-two display segments. Three buttons: up, down, and menu/action. The mandate was a complete thermostat experience for a budget-tier product that still had to feel considered.
At this level of constraint, design becomes editorial.
You cannot show everything, so every character has to earn its place. Every button press has to count. Every hidden layer has to be hidden for a reason.
Navigation stayed flat because hierarchy was too expensive.
Hierarchy is expensive when you have three buttons.
I kept navigation flat and cyclical so users move through settings in a loop rather than drilling into nested menus. The menu/action button serves different roles depending on how it is pressed: a standard press moves through the flow, a long press enters homeowner settings, and a second long press accesses contractor configuration.
That last layer matters.
Contractor settings can break an HVAC configuration if a homeowner stumbles into them. The long-press-to-long-press pattern keeps those settings reachable for installers and out of sight for everyone else, without removing DIY capability.
With flat navigation, order becomes logic.
When navigation is flat, the sequence of settings is not just organizational. It is functional.
Some settings are conditional. Configure a heat pump, and reversing valve direction needs to follow immediately. The order of the flow has to match the logic of the equipment.
I sequenced settings based on dependency logic, with the most common equipment configuration defaulting to the top to reduce installation time.
Testing had to include the whole product.
Usability testing revealed that 23% of users stalled at the three-dot menu button because it did not read as interactive. Auto mode caused hesitation in 30%, confirming the conditional sequencing was doing real work.
Icon testing across five variants landed on a square inside a circle, borrowed from the media stop symbol, and carried forward into later products.
I also observed physical installation: wire tucking, snap-to-baseplate behavior, and clip resistance. Those observations fed directly back into the hardware design.
On a product this constrained, the line between UI and physical design is thin.
When the display goes dark, the system still has to communicate.
Thermostats like Sensi Lite rely on power stealing from the HVAC system. When that power is lost or insufficient, the display cannot function normally.
The question becomes what the system does next and how it tells the user what is happening when the device they would normally check has gone dark.
I was part of the team that worked through that problem: what shuts down, in what order, and how device state gets communicated when the hardware cannot display it.
The answer involved a prioritized shutdown sequence and a notification flow that surfaces device state through the cloud to the mobile app. The user's phone becomes the display when the thermostat cannot be.
That work is the subject of US Patent 12,608,066, awarded April 2026.
The decisions I lost changed how I work.
Two positions did not survive stakeholder review. Post-launch data proved them right.
I wanted fan circulation runtime surfaced higher for homeowner accessibility. It was pushed down in favor of a mobile-first vision and became a recurring customer service issue.
I proposed moving the menu/action button to the opposite side of the display to reduce accidental capacitive taps. Hardware engineering preferred in-line alignment. Accidental taps became a post-launch issue that required a firmware patch.
I do not tell those stories to relitigate them.
I tell them because they changed how I work.
Document the testing rationale. If you are going to lose an argument, lose it with evidence, so when the data comes in, the record is already there.
Designing for guests who will never be onboarded.
The Verdant VX4 and Line Voltage thermostats serve the hospitality market, a fundamentally different design context than residential.
Hotel guests speak different languages, have different levels of technical familiarity, and interact with the thermostat once rather than over years. The interface has to communicate immediately, to anyone, without instruction or support.
There is no onboarding. No help text a guest will read. No long-term learning curve.
The interface works on its own terms or it fails.
Text was not a reliable tool.
The Line Voltage thermostat for the European market pushed this further.
Text was not a reliable tool across the language range the product needed to serve, so I replaced text labels with iconography throughout.
A visual language built entirely around function.
An icon either communicates or it does not, and there is no copy underneath to cover for ambiguity.
The character set had to work from across a hotel room.
Standard 7-segment displays are optimized for manufacturing simplicity, not legibility.
Reading a temperature from across a hotel room is a different problem than reading it up close, and standard segment geometry was not built for that moment.
I started by researching existing standard and non-standard fixed-segment forms, then mapped ideal numerals in Lato, the same typeface used in Sensi Touch 2 where it had already proven itself with users.
Working on transparency paper, I overlaid all digits to find common points and segment cut lines. Then I looked for ways to reduce the total segment count by identifying where segments always appear together and could be treated as one.
That reduction required concessions in less common letterforms like W and Q. User testing confirmed those edge cases were legible enough for the contexts where they appeared.
The testing had to match the room.
The final candidates went to A/B testing using PCB prototypes manufactured by our supplier.
Testing took place in a large room simulating hotel scale, roughly 15 to 20 feet of viewing distance, with participants who had no familiarity with the product.
We varied lighting conditions, displayed random text, and measured both read accuracy and aesthetic preference across versions.
The character set that came out of that process is the subject of a pending US patent. The novel claim sits at the intersection of segment geometry and the reduction method.
Knowledge should not stay trapped in one product line.
The wireless protocol developed for Verdant's hardware found its way into Sensi's room sensor architecture.
Not because the products share an ecosystem, but because the problem was the same and the work already existed.
Expertise built in one context is an asset everywhere the same problem shows up. The organizational cost of siloing it by product line is real.