Compass IoT
Helping traffic operators respond to congestion faster through real-time network intelligence
Compass IoT turns data from cameras, sensors, and connected vehicles into real-time traffic intelligence for governments and road operators globally.
I joined as a product designer to create a congestion management tool that reduces manual monitoring by surfacing congestion events and operational insights in one place.
Compass IoT turns data from cameras, sensors, and connected vehicles into real-time traffic intelligence for governments and road operators globally.
I joined as a product designer to create a congestion management tool that reduces manual monitoring by surfacing congestion events and operational insights in one place.
Compass IoT turns data from cameras, sensors, and connected vehicles into real-time traffic intelligence for governments and road operators globally.
I joined as a product designer to create a congestion management tool that reduces manual monitoring by surfacing congestion events and operational insights in one place.
Revenue
Revenue
+$1.4m
from traffic management centres in Australia alone
of participants desired to use Toastie's AI insights and analytics
User Satisfaction
User Satisfaction
95%
of participants desired to use this in their everyday work
of participants successfully navigated the AI insights with ease
Response Time
Response Time
2x faster
2x faster
incidents detected in 15 min, down from 30–40 minutes
Winner of the Digital Apps and Software Category in 2025
Success Rate
Success Rate
92%
92%
of participants completed incident response without assistance
of participants completed incident response without assistance
Context
Context
Traffic operators are managing increasingly complex road networks with analogue tools
Traffic operators are managing increasingly complex road networks with analogue tools
Induction loops embedded in asphalt, tubes stretched across lanes. Hundreds of CCTV cameras that have to be manually filtered through before a road accident can be confirmed.
This is how most traffic management centres still operate; passive, reactive, and deeply fragmented across tools that were never designed to talk to each other.
Induction loops embedded in asphalt, tubes stretched across lanes. Hundreds of CCTV cameras that have to be manually filtered through before a road accident can be confirmed.
This is how most traffic management centres still operate; passive, reactive, and deeply fragmented across tools that were never designed to talk to each other.
Induction loops embedded in asphalt, tubes stretched across lanes. Hundreds of CCTV cameras that have to be manually filtered through before a road accident can be confirmed.
This is how most traffic management centres still operate; passive, reactive, and deeply fragmented across tools that were never designed to talk to each other.
The Problem
The Problem
Operators spend more time monitoring for congestion than managing it
Operators spend more time monitoring for congestion than managing it
Traffic operators need to identify congestion before it significantly impacts the network. However, critical information is often spread across multiple tools and live feeds, forcing operators to continuously search for emerging issues.
This process creates unnecessary stress and overwhelm during already high-pressure situations and often results in slower decision-making to actively manage network incidents.
Traffic operators need to identify congestion before it significantly impacts the network. However, critical information is often spread across multiple tools and live feeds, forcing operators to continuously search for emerging issues.
This process creates unnecessary stress and overwhelm during already high-pressure situations and often results in slower decision-making to actively manage network incidents.
Traffic operators need to identify congestion before it significantly impacts the network. However, critical information is often spread across multiple tools and live feeds, forcing operators to continuously search for emerging issues.
This process creates unnecessary stress and overwhelm during already high-pressure situations and often results in slower decision-making to actively manage network incidents.
Our Opportunity
Our Opportunity
We had real-time data from cars that were already on the road network
We had real-time data from cars that were already on the road network
Compass IoT has access to a unique source of real-time network information through data directly from cars along almost every road, even in rural and remote Australia.
As vehicles move through the road network, speed, location, and trajectory data can help identify unusual patterns. By comparing current conditions against historical baselines, we can detect emerging queues, congestion, incidents, and network disruptions.
Compass IoT has access to a unique source of real-time network information through data directly from cars along almost every road, even in rural and remote Australia.
As vehicles move through the road network, speed, location, and trajectory data can help identify unusual patterns. By comparing current conditions against historical baselines, we can detect emerging queues, congestion, incidents, and network disruptions.
Compass IoT has access to a unique source of real-time network information through data directly from cars along almost every road, even in rural and remote Australia.
As vehicles move through the road network, speed, location, and trajectory data can help identify unusual patterns. By comparing current conditions against historical baselines, we can detect emerging queues, congestion, incidents, and network disruptions.
The Solution
The Solution
Transforming a rich stream of connected vehicle data into a congestion management experience built around action, not monitoring
Transforming a rich stream of connected vehicle data into a congestion management experience built around action, not monitoring
A single digital experience that takes operators from anomaly detection to incident reporting without leaving one screen.
Incidents are detected and flagged within minutes of when they emerge, with their impact pre-calculated. Live footage is surfaced in context, and reporting is one click away.
A single digital experience that takes operators from anomaly detection to incident reporting without leaving one screen.
Incidents are detected and flagged within minutes of when they emerge, with their impact pre-calculated. Live footage is surfaced in context, and reporting is one click away.
A single digital experience that takes operators from anomaly detection to incident reporting without leaving one screen.
Incidents are detected and flagged within minutes of when they emerge, with their impact pre-calculated. Live footage is surfaced in context, and reporting is one click away.
Timely alerts and immediate breakdowns
Timely alerts and immediate breakdowns
Get notified and immediately see speeds, severity, and queues. No watching required.
Get notified and immediately see speeds, severity, and queues. No watching required.
Trends and insights in one place
Trends and insights in one place
Patterns, comparative graphs, and live camera feeds unified in one view.
Patterns, comparative graphs, and live camera feeds unified in one view.
The Research Behind It
The Research Behind It
In high pressure environments, people just want clarity and guidance
In high pressure environments, people just want clarity and guidance
I ran secondary research and empathy interviews with people working in traffic management and uncovered a deeper look into what their needs actually looked like.
What we found was less about software and more about cognitive load. When decisions affect public safety while the clock is running, every ambiguous data point, every extra tool switch, and every moment spent interpreting rather than acting adds pressure to an already high-stakes situation.
I ran secondary research and empathy interviews with people working in traffic management and uncovered a deeper look into what their needs actually looked like.
What we found was less about software and more about cognitive load. When decisions affect public safety while the clock is running, every ambiguous data point, every extra tool switch, and every moment spent interpreting rather than acting adds pressure to an already high-stakes situation.
I ran secondary research and empathy interviews with people working in traffic management and uncovered a deeper look into what their needs actually looked like.
What we found was less about software and more about cognitive load. When decisions affect public safety while the clock is running, every ambiguous data point, every extra tool switch, and every moment spent interpreting rather than acting adds pressure to an already high-stakes situation.
Key Insights
Key Insights
Lower the cognitive load
Lower the cognitive load
Tap for a clear breakdown of what's happening, why it might be occurring, and what to watch for.
Operators focus on 3-4 clear signals out of hundreds of data points. The remainder just leads to more load and friction instead of value.
Direction leads to confidence
Direction leads to confidence
Export everything you've tracked. Download notes for your doctor, your family, or your own records.
Operators needed guidance and a system that was confident enough in its own output so that they could be confident too.
Design Decisions
Design Decisions
Making the network effortlessly easy to scan
Making the network effortlessly easy to scan
Raw trajectory data can detect speed drops from hundreds of individual data points along a single stretch of road. Surfacing each one as a separate marker would overwhelm the map and contradict everything research told us about cognitive load.
We configured a radius algorithm; alerting only on unique incidents at least 500m apart, a threshold refined through rounds of user testing. We also moved from dot markers to edge colouring, shading road segments by severity so operators could read the network at a glance without interpreting individual pins.
Raw trajectory data can detect speed drops from hundreds of individual data points along a single stretch of road. Surfacing each one as a separate marker would overwhelm the map and contradict everything research told us about cognitive load.
We configured a radius algorithm; alerting only on unique incidents at least 500m apart, a threshold refined through rounds of user testing. We also moved from dot markers to edge colouring, shading road segments by severity so operators could read the network at a glance without interpreting individual pins.
Raw trajectory data can detect speed drops from hundreds of individual data points along a single stretch of road. Surfacing each one as a separate marker would overwhelm the map and contradict everything research told us about cognitive load.
We configured a radius algorithm; alerting only on unique incidents at least 500m apart, a threshold refined through rounds of user testing. We also moved from dot markers to edge colouring, shading road segments by severity so operators could read the network at a glance without interpreting individual pins.
Before
Before
Users felt that the page was very scattered and the overload of information made it hard to see what was important
After
After
We cleaned up the information architecture and moved actionables and resources into each symptom so things felt more holistic
Before
Users felt that the page was very scattered and the overload of information made it hard to see what was important
After
We cleaned up the information architecture and moved actionables and resources into each symptom so things felt more holistic
Design Decisions
Design Decisions
Scaling information to the moment
Scaling information to the moment
Operators don't need the full dashboard the moment an anomaly is detected, they need enough to decide whether to act.
I designed a three-layer progressive disclosure, with each layer triggered by an intentional action.
first, an alert with location and severity; second, a summary insight card with speed deviation, travel time, and queue length; third, the full dashboard with graphs, trajectory data, and live camera feeds.
Operators don't need the full dashboard the moment an anomaly is detected, they need enough to decide whether to act.
I designed a three-layer progressive disclosure, with each layer triggered by an intentional action.
first, an alert with location and severity; second, a summary insight card with speed deviation, travel time, and queue length; third, the full dashboard with graphs, trajectory data, and live camera feeds.
Operators don't need the full dashboard the moment an anomaly is detected, they need enough to decide whether to act.
I designed a three-layer progressive disclosure, with each layer triggered by an intentional action.
first, an alert with location and severity; second, a summary insight card with speed deviation, travel time, and queue length; third, the full dashboard with graphs, trajectory data, and live camera feeds.
Key Learnings
Key Learnings
Trust design instincts even in the unknown
Trust design instincts even in the unknown
I've never worked within such a data-heavy space before, especially not in traffic and road safety. Working extensively with data scientists, the CTO, and a traffic engineer almost led me to create a data-first product instead of a user-first experience.
There were many moments where I had that feeling in my gut that a pattern was broken, or a flow wasn't really meeting the user need but we were limited on how much user testing we could do. As the only designer, I took the initiative to iterate and defend decisions before we did more testing.
Spoiler alert, it paid off!
I've never worked within such a data-heavy space before, especially not in traffic and road safety. Working extensively with data scientists, the CTO, and a traffic engineer almost led me to create a data-first product instead of a user-first experience.
There were many moments where I had that feeling in my gut that a pattern was broken, or a flow wasn't really meeting the user need but we were limited on how much user testing we could do. As the only designer, I took the initiative to iterate and defend decisions before we did more testing.
Spoiler alert, it paid off!
I've never worked within such a data-heavy space before, especially not in traffic and road safety. Working extensively with data scientists, the CTO, and a traffic engineer almost led me to create a data-first product instead of a user-first experience.
There were many moments where I had that feeling in my gut that a pattern was broken, or a flow wasn't really meeting the user need but we were limited on how much user testing we could do. As the only designer, I took the initiative to iterate and defend decisions before we did more testing.
Spoiler alert, it paid off!
Wanna See More?
Wanna See More?
There needs to be a better way for people to understand and manage their health, especially when there is no cure.
This is just a snapshot of the entire project! If you'd like to hear more about the process feel free to reach out to me at chelseavlastalowell@gmail.com
This is just a snapshot of the entire project! If you'd like to hear more about the process feel free to reach out to me at chelseavlastalowell@gmail.com