Contextual AI Insights for Warehouse Operations

Explored AI-assisted predictive layers for an operational control tower, focusing on interpretability, confidence, and decision support rather than automation.

AISystemsDashboard
schedule Jan 2024 - Apr 2024

Problem & Context

Problem Statement

Warehouse managers and supervisors can't get actionable insights on the isolated data provided by Control Tower as it uses static models, limited data, and no recommendations.

To address this, we aim to develop a feature leveraging artificial intelligence, machine learning and advanced analytics to offer actionable insights and recommendations.

How does ML/AI come into the picture?

ML/AI can help answer questions like:

  1. Which products should I re-slot to improve overall efficiency?
  2. If I move three more workers to Zone A, will I get my orders shipped on time?

The Team

Design Team - 1 Interaction Designer (Me), 1 Visual Designer

ML/AI Framework

Finalized Features

Features Overview

  1. AI-driven insights and recommendations appear contextually wherever losses, errors, or anomalies are detected.
  2. Each insight includes a concise explanation along with actionable options to address the issue immediately.
  3. Dashboard widgets can be analyzed individually, generating recommendations specific to the metric or trend shown.
  4. This approach keeps insights relevant, reduces cognitive load, and enables faster decision-making within the existing workflow.
Features Overview UI 1 Features Overview UI 2

Design Parameters

Design Parameters

Use Case : Pharmaceutical

  1. Ryan opens the Control Tower dashboard to review overall warehouse and inventory trends.
  2. He monitors key data points such as inventory levels, inbound and outbound shipments, and compliance status.
  3. An alert appears on the Analyze Card, flagging a potential stock-out risk over the next three days due to rising demand.
  4. Ryan investigates the alert to understand the root cause and projected timeline of the issue.
  5. The system predicts increased demand driven by the flu season, impacting the period from 4th July to 8th July, 2023.
  6. The analysis highlights a high risk of stock-out for essential medicines during this window.
  7. A Recommendation Card suggests corrective actions, including diverting 80% of SKUs to a specific distributor.
  8. The system also shows the predicted impact, indicating that 95% of demand can be met, while warning that remaining stock will require replenishment.
Use Case Pharmaceutical 1 Use Case Pharmaceutical 2