AI, Omnichannel, and the Future of Supply Chain

High-level summary

• Core thesis: Winning omnichannel supply chains are customer-first, data-stitched, and orchestrated end-to-end. Companies must balance automation and labor, rethink processes (not just automate them), and invest in visibility, scenario planning (digital twins), and workforce upskilling to handle volatility, geopolitical disruption, and peaks.
• Practical emphasis: Start from the customer promise (accurate ETA, correct order, quality unboxing), then design network, automation, and AI to deliver that promise cost‑competitively and resiliently.

Frameworks, processes, and playbooks

Customer-back design

• Map the experience from last-mile/consumer backwards into operations.
• Prioritize predictability and quality over absolute speed when warranted.

SKU segmentation + network design playbook

• Identify top-volume/velocity SKUs to position closest to customers.
• Concentrate low-velocity SKUs in regional or central nodes.

Automation vs manual capacity decision flow

1. Assess SKU mix, order volume, seasonality, and growth projections.
2. Choose:
   • Rigid automation for high continuous volume and predictable SKUs.
   • Flexible/manual labor for large periodic peaks.
3. Prefer modular/scalable robotics that can use existing racking where possible.

End-to-end orchestration & visibility playbook

• Stitch origin → carrier → customs → 3PL → domestic transport data.
• Start with dashboards/BI; evolve to AI-driven orchestration and exception management.

Scenario planning / digital twin playbook

• Model the supply network and run “what if” scenarios (e.g., port closure, Strait of Hormuz).
• Use outputs to inform sourcing, inventory location, and contingency plans.

Pilot → Scale governance

• Run pilots with clear ROI metrics.
• Plan for data integration, culture change, and process redesign before scaling.

Contingency sourcing playbook

• Maintain B and C suppliers and nearshoring/friend-shoring options.
• Develop multiple contingency plans (2–3), not just one.

Key metrics, KPIs and targets

• Inventory & network
  • SKU count by node and pool utilization.
  • Percentage of inventory pooled vs segregated.
• Service & customer metrics
  • On-time-in-full (OTIF).
  • Delivery ETA accuracy / predictability.
  • Customer retention / churn (note: customer acquisition cost is often the highest retail expense).
• Operational & financial
  • Forecast accuracy (with ML/AI uplift).
  • Labor utilization during peaks vs idle time.
  • Automation utilization / ROI / payback period (avoid underutilized capex).
  • Exception resolution time and frequency.
  • Number of data sources stitched / end-to-end visibility coverage.
• Strategic forecast
  • By 2030 estimate cited: ~30% of current 3PL services could be done by AI (from referenced report).

Examples, numerical anecdotes, and concrete recommendations

• Organization sizes and SKU ranges:
  • Geodis Americas manages ~20,000 employees across 8 countries.
  • Example clients: one with ~90,000 SKUs; another with ~2,500 SKUs.
• Peak season impacts:
  • Retailers can face 3–10x volume spikes.

Actionable recommendations

• SKU mix and automation
  • Analyze SKU depth/velocity and growth forecasts before investing in expensive automation.
  • Consider modular robotics that retrofit to older facilities.
• Peak management
  • For businesses with concentrated peaks (e.g., two major sales + holiday season), seasonal manual labor may be more economical than full-time automation.
• Visibility & orchestration (ROI today)
  • Automate paperwork/document matching, booking confirmation, and tracking (origin→destination) to gain predictability and operational benefits.
  • Use AI for predictable ETAs and document reconciliation rather than only speculative use cases.
• Forecasting & planning
  • Use ML/AI to improve pattern-based forecasts and labor planning.
  • Deploy digital twins to run scenario planning for geopolitical or supply disruptions.
• AI adoption approach
  • Start with data stitching and a small set of high-value use cases (visibility, exception alerts, forecasting).
  • Redesign processes before automating them.
  • Build governance: data-quality standards, hallucination-risk training, and validation loops.
• Workforce & culture
  • Encourage hands-on use of AI tools for leaders and teams.
  • Re-skill for storytelling, customer engagement, pattern recognition, and critical thinking.
  • Use AI as augmentation to upskill employees (e.g., generate customer briefs, accelerate sales conversations).
• Risk & resilience
  • Maintain alternative sourcing routes (A, B, C) and formalize multi-sourcing contingency plans.
  • Re-architect flows when trade lanes change (nearshoring, friend-shoring, partial onshoring).

Where AI is delivering ROI vs where it’s still maturing

• Real ROI today
  • Information flows: document matching, booking visibility, multimodal tracking.
  • Predictability: earlier, more accurate ETAs that improve downstream operations (sales timing, store sets).
  • Forecasting & labor planning: ML-driven forecasts that convert aspirational numbers into operational plans.
  • Sales enablement & customer brief generation: internal efficiency and revenue opportunity identification.
• Over-hyped / still maturing
  • Full end-to-end autonomous orchestration — data stitching remains the main barrier.
  • Replacing human judgment for complex exceptions, novel events, or negotiation.
  • Point solutions without integration — many pilots have not scaled broadly.

Barriers to AI/automation adoption

• Fragmented data and lack of stitched visibility across carriers, customs, brokers, and 3PLs.
• Automating suboptimal processes instead of redesigning them.
• Culture and mindset resistance; need for change management and new incentives.
• Pilot-to-scale difficulty due to data, process, or governance gaps.
• Model risk: hallucinations, poor training data quality, and failure to validate predictions.
• Automation economics: rigid automation can be underutilized outside peak periods.

New and evolving roles / skills for supply chain professionals

• Storyteller / communicator: translate analytics into actionable narratives for stakeholders.
• Customer-facing / partnership skills: empathetic listening and joint problem-solving.
• Pattern recognition & critical thinking: assess AI outputs, challenge models, and identify root causes.
• AI/tech literacy and applied analytics: use tools, interpret outputs, and set validation metrics.
• Continuous learner mindset: cross-industry learning and hands-on AI upskilling.

Time horizons — what may become autonomous in ~5 years

• Likely to be automated
  • Rule-based exception management and routing adjustments.
  • Improved ML forecasting and labor scheduling.
  • Document processing and many visibility functions.
• Likely to remain human-driven
  • High-judgment decisions, negotiations, and strategic network redesign.
  • Quality control decisions tied to brand experience (unboxing, returns).

Actionable next steps for supply chain leaders

• Inventory & network
  • Segment SKUs by velocity: centralize slow movers, decentralize fast movers near customers.
• Automation & capex decisions
  • Model automation ROI including seasonality and utilization; prefer flexible/modular solutions.
• Data & AI program
  • Start with visibility pilots that stitch key data sources (documents, carrier tracking) and measure improvements in ETA accuracy and exception counts.
  • Roadmap: BI dashboards → ML forecasting → AI orchestration → digital twin scenario planning.
• Workforce & change
  • Launch training that combines AI tool use with storytelling and customer-engagement skills.
  • Run cross-functional playbooks to redesign processes before automating.
• Risk & sourcing
  • Formalize multi-sourcing contingency plans and run digital twin scenarios for critical geographies and lanes.

Notable quotes (paraphrased)

“The last touch in the e‑commerce customer journey is logistics — getting the product to the customer correctly and when you promised it creates brand loyalty.”

“Automation helps throughput but struggles to scale instantaneously for 5x spikes — sometimes seasonal manual labor is the economical choice.”

“We still lack stitched, end-to-end visibility across the trade chain; solving that is foundational to AI orchestration.”

“Don’t automate the way you do things today — rethink what the differentiated services should be in a world of AI.”

Presenters and sources

• Laura Ritchey — President & CEO, Geodis Americas (previously CEO, Radial; COO, Centric Brands; EVP, Victoria’s Secret)
• Eva Ponce — Director of Online Education, MIT Center for Transportation & Logistics; Founder & Director, MIT Omnichannel Supply Chain Lab
• Supporting event team referenced: Elise Jonica and Cara Greeney
• Audience contributors mentioned: Walid Tabbarah and Jorge Gutierrez
• Companies and references: Geodis, Radial, Estée Lauder; referenced industry report forecasting ~30% of 3PL services automated by 2030.