Before
Driver•i
+
HubX
+
External software dependency
Fragmented architecture with integration and compatibility complexity
Project 01
- Role
- Device Product Manager
- Product
- D-810 Multi-Camera Edge-AI Platform
Building an 8-Camera Edge-AI Fleet Platform
Turning customer problems, engineering constraints and product economics into an advanced multi-camera Edge-AI ADAS platform.
Product StrategyEdge AIHardwareConnected Devices
08
One integrated platform
Hardware / Device software / Edge AI
Executive summary
The product challenge
01
Build one integrated multi-camera Edge-AI product.
02
Extend intelligence across up to eight camera streams.
03
Do it without blindly maximizing storage, resolution, hardware cost or deployment complexity.
Customer needs pushed specifications upward. More cameras, better video and more AI created cascading effects across compute, storage, bandwidth, cloud economics, BOM, privacy and installation.
01Context
From a multi-device ecosystem to one product direction.
The existing ecosystem used Driver•i and HubX as separate systems. Driver•i provided AI-powered fleet safety capabilities, while HubX added multi-camera recording without AI analytics across all additional streams.
→
After
D-810
- Integrated multi-camera platform
- Up to eight cameras
- AI across the camera ecosystem
The opportunity was to consolidate important capabilities into a more integrated platform—not merely to add cameras, and not to claim that every architectural element disappeared.
02Customer problems
Four needs. One tightly coupled system.
Each customer request affected decisions elsewhere in the product. The challenge was to understand the need without optimizing one dimension at the expense of the platform.
Problem A
Video quality
Customers wanted better accident and event footage. Some reported difficulty identifying details such as number plates from existing footage.
Quality ↔ file size, storage, bandwidth and cloud cost
Problem B
Storage
Some customers requested longer retention. With eight cameras, scaling the previous storage model would materially increase storage requirements and hardware cost.
Retention ↔ SSD sizing, BOM and configuration
Problem C
Privacy
Some customers—particularly in Europe—had strict expectations around retaining identifiable video involving drivers and people around vehicles.
Customer access ↔ privacy constraints
Problem D
Installation
More side and rear cameras introduced additional positioning, calibration and installation work across every vehicle deployment.
Coverage ↔ installation time and deployment cost
03My role
Product decision-making across the device system.
“As Device Product Manager, I owned the hardware and device-software requirements and acted as the product decision maker across the major trade-offs.”
01 / Customer & market
- Customer needs
- Market research
- Field feedback
02 / Product definition
- Hardware requirements
- Device software requirements
- Camera capabilities
- Configuration
- Product limitations
03 / Product economics & scale
- BOM
- Deployment complexity
- Commercial scalability
- Lifecycle considerations
My role was to identify which capabilities genuinely created customer value, then make product decisions that balanced technical ambition with commercial and operational scalability.
04Research & storage decision
Do customers actually need more storage?
Decisions combined customer feedback, usage analysis, field feedback, engineering analysis, installer considerations, and privacy and market requirements.
“More storage was frequently requested, but usage analysis showed that most requested video was retrieved relatively soon after an event.”
This challenged the assumption that maximizing retention automatically created customer value. It reframed storage as a product-economics decision rather than a specification contest.
Customer request“More retention”
Usage analysis
Low usage of older stored video
Product questionIncrease cost for a rarely used capability?
Decision principleOptimize for demonstrated value
Implication
SSD sizing
Implication
BOM
Implication
Configuration
Implication
Privacy and long-term economics
05Core product trade-offs
More capability moved the whole system.
Better video and AI across additional cameras increased customer value—but also affected compute, storage, bandwidth, recurring cloud economics and device cost.
Decision A / Video quality
Better evidence without uncontrolled cost.
Video quality was treated as a system-level product decision, not an isolated camera specification.
Higher video quality
Larger files
More storage
More bandwidth
Higher recurring infrastructure cost
“The right video quality for the customer use case at a sustainable product cost.”
Decision B / AI across the ecosystem
From recording more views to creating more intelligence.
Extending AI across the camera ecosystem introduced complexity and cost. It also represented the core product differentiation.
Traditional multi-camera recording
Record more views.
D-810 vision
Use additional views to contribute intelligence, event context and early warning capability.
06Configurability
Configuration instead of one-size-fits-all.
Customers prioritized video quality, retention, privacy, camera configuration and deployment cost differently. Maximizing every specification in one fixed configuration would create a more expensive, inflexible product.
More engineering complexity upfront
More configurable platform
Better customer and market fit
Greater scalability across deployments
07Deployment economics
Installation is part of the product.
At fleet scale, small increases in per-vehicle installation time become meaningful deployment costs. Additional cameras and calibration had to inform product decisions—not remain only an operations concern.
“A physical product is not complete when it leaves engineering. It is complete when it can be deployed reliably at scale.”
08Decision framework
Every choice moved the whole system.
The product decision sat at the intersection of customer value, technical capability, economics, privacy and deployability.
01AI capability
02Video quality
03Storage
04Bandwidth
Center
Product decision
05Privacy
06BOM
07Installation complexity
08Customer value
09Outcome & lessons
A more integrated multi-camera product direction.
01
Integrated multi-camera platform
02
Up to eight camera streams
03
AI analytics across the camera ecosystem
04
Improved video-quality capability
05
Configurable customer and market behavior
06
More integrated product architecture
What I learned
Optimization—not maximization—is the product work.
Feature requests are not automatically requirements.
Hardware decisions create recurring software and cloud economics.
Configurability can justify higher engineering complexity.
Great hardware product management is optimization, not maximization.
“The objective was not to maximize every specification. It was to find the right balance of customer value, technical capability, economics and operational scalability.”