AI Supply Chain Talks That Actually Change Operations

A lot of “AI in logistics” conference talks are really product demos with a nicer title. They’re easy to spot: vague claims, no baseline metrics, and a quiet slide that says “results may vary.” Most companies get this wrong—because the audience doesn’t need more AI hype. They need proof that intelligent operations and resilient supply chains can be built, measured, and improved.

That’s why the ARC Industry Forum 2026 call for speakers matters more than it looks at first glance. The forum is asking for vendor-neutral, real-world case studies and strategic frameworks—the kind of sessions that help an operations leader walk back into Monday’s standup with a plan, not a buzzword.

This post is part of our “AI in Supply Chain & Procurement” series, where we focus on practical AI: forecasting, supplier risk, transportation planning, warehouse execution, and the messy integration work that makes it real.

Why intelligent operations needs AI (and why resilience depends on it)

Intelligent operations is the ability to sense change, decide fast, and execute reliably across planning and execution. AI is the only toolset that scales that loop.

Resilience used to mean holding extra inventory and having backup carriers. That still helps, but it’s expensive and slow. In 2026 planning cycles, resilience is increasingly about:

• Detecting disruption early (supplier delays, port congestion, capacity drops)
• Quantifying tradeoffs (service vs cost vs emissions vs working capital)
• Executing corrective actions (replan, reroute, reallocate, expedite) without chaos

AI fits because logistics and procurement generate high-volume signals—orders, shipments, scans, exceptions, lead times, invoices. The real issue isn’t “Can AI do it?” It’s “Can we operationalize it across systems and teams?” That’s exactly the gap a strong conference session can close.

A myth worth killing: “Resilience is a strategy document”

Resilience isn’t a slide deck. It’s a set of capabilities that show up in daily execution:

1. Visibility: near-real-time understanding of what’s happening
2. Prediction: what’s likely to happen next (ETAs, demand shifts, supplier performance)
3. Decisioning: recommended actions with constraints and consequences
4. Automation: workflows that make the action happen (or at least route approvals fast)

If your AI initiative doesn’t improve at least two of these in a measurable way, it’ll stall.

What ARC 2026 is really asking for in speaker proposals

The RSS post is a call for speakers focused on intelligent operations and resilient supply chains, with themes like:

• AI, analytics, automation, and connected intelligence
• Resilience under shifting geopolitics (tariffs, regulation, policy)
• Knowledge transfer inside enterprise systems

The explicit standard is important: vendor-neutral, educational, senior-exec relevant. That’s a higher bar than many events.

If you’re considering submitting, treat it like you’re writing a case memo for your board:

• What problem did you have?
• What did you change (process + tech + governance)?
• What improved (numbers, reliability, risk, speed)?
• What would you do differently next time?

The sessions people remember are “how we did it” sessions

I’ve found the best logistics AI talks share uncomfortable details:

• Data that wasn’t clean (and what you did anyway)
• A model that performed well in a lab and failed in a DC
• The one KPI that mattered more than accuracy
• The workflow change that actually delivered ROI

That’s what peers come to learn.

High-impact AI supply chain topics that fit ARC’s themes

If you want to “drive real change,” anchor your talk around a measurable loop: signal → prediction → decision → execution → learning. Below are topics that map cleanly to intelligent operations and resilience—and make for strong proposals.

1) Demand forecasting that survives promotions, substitutions, and chaos

Most forecasting discussions fixate on algorithms. The real win is making forecasting usable by merchandising, planning, and procurement.

Strong angles for a session:

• Forecast value-add (FVA): how you measured which steps improved forecast vs added noise
• Managing promotion lift, cannibalization, and price elasticity without overfitting
• Using AI to recommend inventory positioning (not just predicting demand)

What “real change” looks like:

• Fewer stockouts on A items without inflating total inventory
• Lower expedite spend because procurement lead times are planned realistically

2) Supplier risk scoring that procurement teams actually trust

Procurement leaders are tired of black-box “risk scores” that don’t change decisions.

A talk that lands well:

• Combining supplier OTIF, quality escapes, financial signals, and geopolitical exposure into an actionable model
• Tying risk outputs to sourcing policy: dual-source thresholds, safety stock rules, contract clauses
• Establishing a human override process (and tracking when overrides were right)

Snippet-worthy line for your abstract:

“A risk model is only useful if it changes a purchase order before it becomes an expedite.”

3) Transportation planning with AI: better decisions, not just better ETAs

Predictive ETAs are table stakes. The next step is decisioning.

Topics that earn attention:

• AI-assisted mode selection and dynamic routing under capacity constraints
• Exception management that reduces planner workload (fewer false alarms)
• Using AI to predict accessorial risk (detention, demurrage, redelivery) and prevent it

Practical metric ideas to share:

• Tender acceptance rate
• Cost per shipment (normalized for fuel)
• On-time delivery at request date (not just promise date)

4) Warehouse execution: where AI meets labor reality

Warehouses are where “AI transformation” either becomes visible—or gets rejected.

A strong case study might cover:

• Slotting optimization tied to travel time and congestion
• Labor planning that predicts workload by zone and time block
• Computer vision for damage, dimensioning, or safety (with a clear governance model)

If you can show a before/after of:

• Lines per labor hour
• Missed cutoffs
• Safety incidents

…you’ll have the room.

5) Knowledge transfer: the quiet bottleneck in intelligent operations

ARC’s theme on knowledge transfer is underrated. Many supply chains are held together by a few people who “just know” how things work.

A talk here can be extremely practical:

• Turning tribal knowledge into standard work + decision trees
• Capturing exceptions and resolutions inside TMS/WMS/ERP workflows
• Using internal copilots safely: approved playbooks, retrieval-based answers, and audit trails

This matters because resilience fails when expertise is a single point of failure.

What a “vendor-neutral” AI case study looks like (a simple template)

The best proposals follow a structure that makes it impossible to hide the hard parts. If you’re writing an abstract (150–250 words), this outline works.

1) Baseline: the operational pain in one paragraph

Examples:

• “We were missing OTIF targets due to late supplier shipments and reactive expediting.”
• “Transportation planners spent 40% of their time on exceptions with low signal.”

2) Approach: data + workflow + governance

Be specific:

• Data sources used (ERP, WMS, TMS, EDI, IoT, supplier portals)
• Model type at a high level (forecasting, classification, anomaly detection, optimization)
• Where it sits in the workflow (who sees the recommendation, who approves, what triggers it)

3) Results: 3–5 KPIs with time horizon

If you can’t share exact numbers, share ranges and timeframes:

• “Within 90 days, exception volume dropped by ~25–35%.”
• “Within two quarters, expedite shipments fell by ~10–15% while service held.”

4) Lessons: what broke, what you changed

This is the credibility engine.

• What data surprised you?
• Where did humans ignore the model?
• What governance avoided bad automation?

People also ask (and what I’d say on stage)

“Should we start with generative AI or predictive AI in supply chain?”

Start where you can measure operational lift quickly. In most networks, that’s predictive/optimization (forecasting, ETA, risk, replenishment). Use generative AI for knowledge retrieval, SOP support, and analytics assistance—then expand once controls are in place.

“What’s the biggest blocker to AI in logistics?”

Workflow adoption. If the output doesn’t show up inside the tools planners and buyers use every day, it becomes a dashboard nobody opens.

“How do we keep AI from becoming a compliance risk?”

Put governance in the design: data lineage, role-based access, audit trails, and clear rules for when automation is allowed vs when human approval is required.

A practical next step: turn your AI initiative into a speaker proposal

The ARC Industry Forum 2026 call for speakers is an opportunity to do two things at once:

• Share what actually worked in intelligent operations and resilient supply chains
• Pressure-test your own program narrative—because if you can’t explain it clearly, you probably can’t scale it

If you’re a supply chain, logistics, operations, or procurement leader, I’d take a hard stance here: submit a proposal only if you’re willing to teach the room something you learned the hard way. That’s what peers will remember, and it’s what builds credibility.

As our “AI in Supply Chain & Procurement” series keeps stressing, the winners in 2026 won’t be the teams with the fanciest models. They’ll be the teams that can turn predictions into decisions—and decisions into executed work—every single day.

What’s the AI-driven operational loop in your organization that’s finally working end to end—and what did it take to get there?