Blog #24. AI. It’s time.
A while back, recruitment guru Katherine Biggelaar prompted me to write about AI in Supply Chain, specifically how it should be used in IBP/S&OP. Sorry Kat - it's taken me ages to respond, but not for the reason you might think.
The problem wasn't figuring out where to use AI. The problem was figuring out what not to say about it.
It's already changed the way many of us do things. I love it and use it prolifically. But like any adoption curve, it's scary because it’s new and unfamiliar. People say we are a few years away from trusting it the way the hype suggests we should, but I’m not so sure.
Here's a truth though: all your staff are using it flat out. You probably are too.
That email you sent about not uploading confidential information to ChatGPT? They read it. Then uploaded their forecast spreadsheet anyway because it helped them spot the error in three minutes instead of three hours.
AI bans don't work. Companies issue policies, employees ignore them, and everyone pretends the problem is solved.
Your Ban Isn't Working
According to a 2025 Anagram Security report, 45% of employees are using AI tools that their companies have banned. On personal accounts. With company data. And 58% admit to pasting sensitive information - client records, financial data, internal documents - straight into ChatGPT.
That email you sent about not uploading confidential information? They read it. Then uploaded their forecast spreadsheet anyway because it helped them spot the error in three minutes instead of three hours.
AI bans don't work. Companies issue policies, employees ignore them, and everyone pretends the problem is solved.
Cisco's 2024 Data Privacy Benchmark Study found that 27% of organisations have banned generative AI tools over privacy and data security fears. Yet 48% of employees admit to entering non-public company information into these tools anyway.
Your Supply Chain folk aren't reckless or stupid. They're buried under work, and AI genuinely helps. When leadership hasn't provided approved tools or clear guidance, people grab whatever works.
The Real Problem
Yes, the fear is real. Proprietary information, customer data, strategic plans - all potentially ending up in AI training datasets or exposed to competitors.
Remember when Samsung employees uploaded confidential source code to ChatGPT in 2023? Three separate incidents in under a month - source code, meeting notes, chip testing data - all ingested into OpenAI's systems. That's the scenario keeping your CIO awake at night.
But here's another truth: your people have been "gifting away" company information for decades.
Confidential emails to personal Gmail. Strategic conversations in airport lounges. Presentations left in taxis. Internal gossip dropped on LinkedIn. The contractor who walked out with your customer database on a USB stick.
AI didn't create this risk. It just made you pay attention to it.
Katherine didn't ask, "Should we use AI?" Her question was: "How should supply chains use it?"
AI can absolutely help with planning work. Pattern spotting in demand data. Finding forecast errors. Drafting scenario analyses. Challenging dodgy assumptions.
But if you're nervous about uploading confidential information - don't. Start somewhere safer.
Start Anonymous, Start Generic
Here's the thing: some of the most valuable AI applications in IBP don't require you to upload a single confidential number.
Assumptions and Risks & Opportunities are perfect examples. These are largely generic for your industry. You're not uploading your sales figures or customer names - you're asking AI to help you think more rigorously about market conditions, supplier risks, competitive dynamics, and demand drivers.
"I'm a demand planner in [industry]. What assumptions should I be challenging about raw material availability over the next two quarters? Consider supplier concentration risk, geopolitical factors, and lead time variability. Push back on assumptions that might have been true two years ago but aren't now."
"Act as a devil's advocate for my S&OP process. I work in [industry], and we're planning for seasonal peak demand over the traditional December retail peak. What risks do companies in this sector typically underestimate? What opportunities do they miss because they're focused on the obvious ones? Be specific and contrarian."
"Help me structure a scenario analysis for currency fluctuation impact on a [industry] supply chain. I need three scenarios - baseline, adverse, and severe - with the key planning assumptions that would change under each. Focus on the second and third-order effects that planning teams often miss."
Notice the detail in these prompts. There's an art to prompt engineering - the difference between a useful response and a useless one often comes down to how you ask. More on that another time.
None of this exposes your proprietary data. All of it sharpens your planning.
This is only the start. There's much more AI can do once you've built confidence and established sensible guardrails - but that's for next time.
The Bottom Line
Your people are using AI anyway, on tools you don't control, in ways you can't see.
The answer isn't banning what you can't enforce. It's providing guidance on what's safe, what's useful, and what's off-limits.
Your call.
Are your people using AI despite the policies? What's actually happening in your world?
Want to talk about using AI sensibly in IBP? Find me at www.planninglab.co.nz
#IBP #S&OP #BusinessPlanning #ArtificialIntelligence #SupplyChain #DemandPlanning