Practical AI for Supply Chain Execution: What Actually Works in Real Operations
There’s no shortage of conversation around AI in supply chain right now.
But most of it still lives in one of two places:
- Too abstract to be useful
- Too technical to apply
This March 25 virtual session with ASCM Twin Cities takes a different approach.
Instead of talking about what AI could do, Richard Lebovitz, Founder and Chief Strategy Officer of LeanDNA, focuses on what’s already working inside real manufacturing environments.
AI isn’t the problem. Execution is.
One of the clearest themes from the discussion is this:
Supply chain planning has improved. Execution hasn’t kept up.
Most organizations already have forecasting tools, ERP systems, and reporting dashboards, but they still struggle to answer simple, high-impact questions fast enough:
- Where are we actually at risk?
- What should we act on first?
- What will impact production if we don’t act now?
That shift - from planning to execution - is where AI is starting to deliver real value.
Why AI is accelerating now
AI didn’t just show up overnight. What’s changed is the environment around it.
Technology has matured.
AI models can now understand supply chain context - things like order policies, demand variability, and supplier performance.
Data is finally usable.
Between ERP systems and supplier collaboration tools, companies now have access to the historical and real-time data AI needs.
Where AI actually works
A lot of companies start in the wrong place with big, expensive transformations or abstract “AI strategies.”
The teams seeing results are doing something much simpler: applying AI to specific execution problems.
This typically shows up in a few ways:
- Identifying shortages earlier - before they impact production
- Prioritizing which POs or suppliers actually matter
- Reducing the noise coming out of ERP systems
- Automating decisions that used to require manual analysis
From insight to action
Most systems today already generate insights.
What they don’t do well is:
- Prioritize actions
- Assign ownership
- Track outcomes
- Learn over time
The model that’s emerging is a loop:
Optimize → Execute → Learn → Improve
What’s next
The next phase isn’t just better analytics.
It’s systems that:
- Guide workflows
- Handle ambiguity
- Coordinate across teams and suppliers
Where to start
If there’s one takeaway from the session, it’s this: Don’t start with AI. Start with a problem.
Pick something like:
- Shortage visibility
- Inventory imbalance
- Supplier risk
Focus on a clear outcome. Move quickly. Build from there.
Because the companies getting value from AI right now aren’t the ones with the biggest strategies.
They’re the ones applying it where it actually matters: inside the day-to-day decisions that drive production.
If you want to learn more about LeanDNA's AI capabilities for discrete manufacturers or speak with one of our supply chain strategists, let us know.