AI Decision Intelligence Platform
AI Decision Intelligence Platform
AI Decision Intelligence Platform
Role: Lead Product Designer
Scope: UX Strategy, System Architecture, Interaction design, Visual Identity
Role: Lead Product Designer
Scope: UX Strategy, System Architecture, Interaction design, Visual Identity
Team: Product, Engineering
Stakeholders: Warehouse Managers & Operational Teams
Team: Product, Engineering
Stakeholders: Warehouse Managers & Operational Teams
OVERVIEW
OVERVIEW
How might we help warehouse teams turn complex operational data into clear decisions that improve system efficiency and throughput?
I led design for Decision Intelligence — Honeywell's AI-powered upgrade to their Momentum warehouse platform. Working alongside the product owner to shape strategy, I owned the full design direction, made the core UX decisions, and crafted the visual identity that positioned this feature within Honeywell's broader ecosystem. The mandate was clear: make AI feel actionable, not overwhelming, for warehouse teams operating in real time.
I led design for Decision Intelligence — Honeywell's AI-powered upgrade to their Momentum platform. I co-owned the strategy with the product owner, drove all design direction, and crafted the visual identity. The goal: make AI feel actionable, not overwhelming.
IMPACT
IMPACT
IMPACT
40% reduction in manual updates through AI-driven automation within the first quarter—improving shift transition efficiency by an average of 26%.
40% reduction in manual updates through AI-driven automation within the first quarter—improving shift transition efficiency by an average of 26%.
40% fewer manual updates. 26% faster shift transitions—with AI automation.
CHALLENGE
Warehouse teams were drowning in data but lacked clear decisions.
Warehouse managers relied on AS/RS systems that generated massive operational data, but the insights were difficult to interpret in real time. Managers often had to manually adjust allocations, pacing, and storage zones during shifts—introducing delays and guesswork into systems designed to be automated.
Small inefficiencies quickly compounded across the warehouse floor. Manual adjustments could take 20–30 minutes per shift, forecasting errors increased dwell and travel times, and nearly 15% of orders required last-minute intervention.
The core problem wasn’t a lack of data.
It was that the system couldn’t translate data into clear operational decisions.
Warehouse managers relied on AS/RS systems that generated massive operational data, but the insights were difficult to interpret in real time. Managers often had to manually adjust allocations, pacing, and storage zones during shifts—introducing delays and guesswork into systems designed to be automated.
Small inefficiencies quickly compounded across the warehouse floor. Manual adjustments could take 20–30 minutes per shift, forecasting errors increased dwell and travel times, and nearly 15% of orders required last-minute intervention.
The core problem wasn’t a lack of data.
It was that the system couldn’t translate data into clear operational decisions.
Warehouse managers had mountains of data but no clear way to act on it. Manual adjustments took 20–30 minutes per shift, forecasting errors slowed fulfillment, and 15% of orders needed last-minute intervention. The problem wasn't data. It was decisions.
Strategic Opportunity
Momentum already captured operational data across AS/RS systems but, the platform lacked a clear way to translate this data into actionable decisions. Decision Intelligence was introduced to bridge this gap.
Momentum already captured operational data across AS/RS systems but, the platform lacked a clear way to translate this data into actionable decisions. Decision Intelligence was introduced to bridge this gap.
Momentum already captured operational data across AS/RS systems but, the platform lacked a clear way to translate this data into actionable decisions. Decision Intelligence was introduced to bridge this gap.
Warehouse teams didn’t need more data.
They needed clearer decisions.
Warehouse teams didn’t need more data.
They needed clearer decisions.
DESIGN STRATEGY
Making AI actionable at a glance
To bring clarity to a complex system, we designed around how warehouse teams actually work—what they need to adjust, monitor, and trust in real time. We structured the interface into three core layers:
1. Smart Configuration
Simplified Operators needed more than control—they needed clarity. We used progressive disclosure and intuitive grouping to let users adjust AI parameters without getting lost in technical jargon.
2. Monitoring That Builds Trust
We added safety rails and fallback protocols. Alerts were designed to keep teams proactive—not reactive—with human-readable, high-signal messaging.
3. Planning with Visual Intelligence
AS/RS Activity Profile became the heartbeat of the tool, showing SKU popularity tiers at a glance. This helped teams optimize storage and container planning based on real-time and forecasted movement.
To bring clarity to a complex system, we designed around how warehouse teams actually work—what they need to adjust, monitor, and trust in real time. We structured the interface into three core layers:
1. Smart Configuration
Simplified Operators needed more than control—they needed clarity. We used progressive disclosure and intuitive grouping to let users adjust AI parameters without getting lost in technical jargon.
I structured the experience around three layers — and I defined them.
My north star: keep it minimal enough for a warehouse floor, not a boardroom.
Smart Configuration — AI controls grouped simply. No jargon. Just clarity. Trustworthy Monitoring — Proactive alerts and safety rails teams could rely on. Visual Planning — SKU popularity visualized so managers could optimize storage in real time.
2. Monitoring That Builds Trust
We added safety rails and fallback protocols. Alerts were designed to keep teams proactive—not reactive—with human-readable, high-signal messaging.
3. Planning with Visual Intelligence
AS/RS Activity Profile became the heartbeat of the tool, showing SKU popularity tiers at a glance. This helped teams optimize storage and container planning based on real-time and forecasted movement.
Together, these layers created a single source of truth—one that helps teams move faster and make smarter decisions with confidence. These three layers weren't handed to me — I defined them. My north star was keeping the interface minimal enough for a warehouse floor, not a boardroom.
Together, these layers created a single source of truth—one that helps teams move faster and make smarter decisions with confidence. These three layers weren't handed to me — I defined them. My north star was keeping the interface minimal enough for a warehouse floor, not a boardroom.
See the design process
See the design process
Collaboration and Constraints
Designing Decision Intelligence required close collaboration across product, engineering, and operations teams to ensure the system worked within real warehouse environments.
• Engineering: Worked with backend teams to define prediction thresholds, monitoring signals, and fallback conditions for AI models.
• Operations: Validated decision signals and configuration controls with warehouse operators to ensure recommendations aligned with real shift workflows.
• Product Leadership: Helped define how Decision Intelligence would integrate with the broader Momentum platform and complement existing reporting tools.
I worked closely with engineering to define AI thresholds and fallback logic, validated decisions with warehouse operators, and aligned with product leadership on platform integration. Cross-functional by design — from day one.
How Decision Intelligence Actually works
How Decision Intelligence Actually works
Decision Intelligence continuously evaluates warehouse signals to determine when automation should assist operators and when manual control should remain in place.
STEP ONE
STEP ONE
STEP ONE
Monitor Signals
Decision Intelligence monitors warehouse signals like dwell time, space usage, and SKU velocity.
STEP TWO
STEP TWO
STEP TWO
Evaluate Conditions
The system evaluates these signals against predefined thresholds and predictive models.
STEP THREE
STEP THREE
STEP THREE
Adjust or Defer
If conditions are met, AI predictions automatically adjust. If confidence is low, operators maintain control.
This model ensured automation remained adaptive while preserving operator control during uncertain conditions.
SOLUTION
A smart, responsive dashboard that drives faster decisions and smoother ops
We launched a streamlined automation interface called Decision Intelligence—designed to adapt dynamically to the evolving needs of each warehouse shift.
Key features included:
A centralized configuration UI for AS/RS systems, offering intuitive control over every subsystem
Real-time pacing metrics to monitor throughput and identify slowdowns before they escalated
AI-backed auto-allocation of storage zones, bins, and dwell times based on shifting inventory patterns
Clear protocols for overrides and exception handling, giving managers control in edge cases
We also introduced the AS/RS Activity Profile—a data visualization and planning tool that used SKU ranking curves to optimize container profiles. This helped classify inventory into popularity tiers, informing smarter storage decisions and reducing retrieval time.
The system didn't just automate — it learned. And the numbers showed it.
We delivered Decision Intelligence—a streamlined interface with centralized AS/RS configuration, real-time pacing metrics, and AI-powered zone and bin allocation.
The AS/RS Activity Profile enabled smarter storage decisions based on SKU ranking, while override protocols ensured control. The system saved time and continuously adapted to shifting operational needs.
Platform Impact
Design Decision: Build on what exists, don't start from scratch.
I had already crafted the component library for Honeywell's Warehouse Reporting system. When Decision Intelligence came along, I made the deliberate call to extend those components rather than reinvent them. That decision kept the experience consistent for users who moved between tools, and cut design-to-dev handoff time significantly. That's not just good UX — that's good product thinking.
I'd already built the component library for Honeywell's Warehouse Reporting system. I made the deliberate call to extend those components here — keeping the experience consistent across tools and cutting handoff time. That's not just good UX. That's good product thinking.
Key Product Design Decisions
Key Product Design Decisions
Why separate monitoring dashboards from configuration?
Why we chose operator-controlled toggles instead of fully automated AI
Why include fallback logic for advanced predictive models?
Why keep the main configuration interface so minimal?
RESULTS
RESULTS
RESULTS
The decision to keep configuration minimal, surface only high-signal alerts, and build on existing patterns didn't happen by accident — it was the strategy. Here's what it produced:
The strategy was intentional. Here's what it produced:
40%
40%
Reduction in manual system updates
Reduction in manual system updates
26%
26%
improvement in shift transition efficiency
improvement in shift transition efficiency
• Reduced operational guesswork by surfacing AI-driven recommendations
• Enabled warehouse teams to monitor and adjust system behavior in real time
• Reduced operational guesswork by surfacing AI-driven recommendations
• Enabled warehouse teams to monitor and adjust system behavior in real time
• Reduced operational guesswork by surfacing AI-driven recommendations
• Enabled warehouse teams to monitor and adjust system behavior in real time
