There’s a lot of fluffy marketing talk about AI in ERP right now, and I get why. I’d probably be saying the same things if I sat on the other side of the table. But twenty years of buying software as a CFO, and now over a decade of building and selling it, taught me to ask one thing before anything else: does this AI feature actually help someone today, or is it next year’s roadmap slide dressed up for a demo?
That question matters more for beverage producers than almost anyone else I work with. Your ERP touches your TTB filings, batch records, and your margin on that seasonal release that’s only made once a year. Get the “is this real” question wrong, and it costs you. So here’s my honest read on where AI in ERP actually stands, what I’d pay for today, and what I’d want answered before I believed anything else.
Why Is Every ERP Vendor Suddenly an AI Company?
Analysts have started calling this out publicly, which tells you something. Gartner has used the term “agent washing” to describe vendors rebranding ordinary automation as agentic AI without adding real capability. The firm expects a significant share of agentic AI projects to be canceled once costs and unclear returns catch up with the promises. Forrester’s 2026 predictions describe next year as the point where the AI hype period ends and the pressure to show measurable results finally lands.
None of that means AI in ERP is fake. It means the gap between what gets demoed and what gets deployed is wider than most buyers realize, and beverage producers don’t have the budget or the time to fund somebody else’s experiment.
The Data Foundation Nobody Wants to Talk About
Vendors skip past this part fast, but it’s the one thing that actually matters: AI is only as good as the data underneath it. I’ve watched teams dump AI on top of a mess of spreadsheets and disconnected point solutions hoping it’ll sort itself out.
Spoiler: It doesn’t. It just makes the wrong number sound more official.
The alcohol industry makes this harder than most. TTB compliance alone means daily operational records, monthly reports, and batch genealogy must hold up to an audit. Most industries don’t deal with anything close to that.
Now add seasonality. A wheat beer that sells out by August, a rosé timed to one harvest a year, or an RTD spike the week before a holiday. Try forecasting any of that, even with clean data, and you’ll feel the difficulty firsthand.
Layer AI onto a Frankenstack of disconnected systems, duct-taped together with brittle data pipelines, and you haven’t solved the problem. You’ve automated the guessing.
Where Is AI in ERP Actually Earning Its Keep?
Honestly? Fewer places than the marketing suggests, but the ones that exist are real. Financial anomaly detection is one of them. It catches an unusual invoice or a duplicate payment before it posts, rather than after your controller stumbles upon it during close.
AI-assisted drafting takes a first pass at the write-ups nobody has time for — the reports, the notes, the paperwork that piles up — so a person can edit instead of starting from scratch. And forecasting gets sharper over time because it’s learning from your actual transaction history, not some generic industry curve.
I’ve seen this play out with real producers, not hypothetical ones. Before AI was part of the conversation, TX Whiskey used Crafted ERP’s approach to production planning to cut a process that once took 40 hours a week to about 30 minutes, while scaling from 2,000 to 30,000 barrels a year. That’s not an AI result. That’s clean data and the right automation doing the heavy lifting on their own, which is exactly the foundation AI needs to be useful at all. Skip that step, and AI has nothing solid to stand on.
What Beverage Producers Are Still Missing
I wouldn’t buy an ERP platform today on the promise that it’ll run your compliance filings or your production schedule without a person in the loop. That’s not actually the goal, and it shouldn’t be. Your team’s judgment on a batch that’s behaving strangely or a filing that doesn’t look right is the thing you’re paying for. What I’d buy AI for is to make that team faster: a force multiplier that takes the repetitive work off their plate so their judgment can go further, not a replacement for it.
NetSuite is actively rolling out Ask Oracle and NetSuite Next, its AI-native ERP experience, in the U.S. and Canada, with more capabilities and regions on the way. Our product team is finishing the work to bring Crafted fully into that experience. It isn’t there yet, and I’d rather tell you that than let it imply otherwise. “Coming” and “here” are different purchasing decisions. Buy for the platform’s trajectory, not for a feature you can’t use today.
What Should You Ask in Your Next AI Demo?
Start with the basics: is this feature live right now, or still on the roadmap? Either way, get a real release date, not a vague timeline. Then make them prove it. Ask to see actual data, not a polished demo sample, and watch it run live.
Find out where your data is processed and whether it’s used to train models shared with other companies, too. And ask the hardest question last: when the AI gets it wrong, does a human catch it, or does it just act on its own? Vague answers to any of that tell you everything you need to know.
Built in Oracle NetSuite, Built for The Beverage Industry
This is exactly why we built Crafted ERP in Oracle NetSuite rather than layering software on top of a generic platform. NetSuite’s AI capabilities, like Bill Capture and Financial Exception Management, sit inside the same system that already holds your production, compliance, and financial records, so the AI has real, unified data to work with instead of scraps pulled from five different tools.
And because Crafted starts every implementation by getting your data clean and connected first, you aren’t hoping AI fixes a messy foundation later. You’re building on a clean one from day one.
The Bottom Line
AI in ERP is neither a trap nor a miracle. It’s a set of genuinely useful tools that only work as well as the data you feed them, and beverage producers can’t afford to gamble that distinction away on a vendor’s roadmap slide. If you want an honest look at where your own data stands and what AI could actually do for your operation once it does, let’s talk.
About Jeremy King
Jeremy King is CEO and co-founder of Doozy Solutions, the team behind Crafted ERP. His background as a CFO and COO gives him a hands-on perspective on the operational and financial challenges facing manufacturing, technology, and craft beverage businesses. His writing on technology and cloud computing has appeared in the Denver Business Journal, CoBiz Magazine, and Accounting Today. Jeremy holds a bachelor’s degree in economics from UCLA.
