How Vincero Used AI to Rebuild Their Planning Process

Vincero wasn't operating in chaos. 

They had built something sophisticated: interconnected Google Sheets tracking 600+ active SKUs (over 1,000 with engraving variations) across four product categories (watches, jewelry, eyewear, and carry accessories) with formulas calculating velocity, managing inventory positions, and tracking outstanding POs from multiple suppliers.

But sophistication created its own problems. 

Sean describes a system where sheets sat on top of sheets on top of sheets, all connected in precise ways. Adjust one formula incorrectly and everything would break. 

This made the system rigid and difficult to change, which was a problem because the primary thing demand planning should enable is scenario planning.

The complexity compounded across several dimensions:

This created three constraints that compounded over time:

1. Marketing couldn't get answers fast enough: When Aaron (as CMO) asked what would happen if they shifted budget from one SKU to another, Sean would need hours to manually adjust formulas across multiple sheets. By the time he had an answer, the opportunity had passed.

2. Every rule had exceptions that broke the system: Lead times would change. Suppliers would shift. The business would evolve. But the spreadsheets couldn't flex without risk of breaking the entire structure. Aaron explains they needed something that wasn't fixed, something more pliable that could adapt to where the business was going while still handling where it was today.

3. Knowledge was siloed in one person: Aaron would ask Sean about stock levels for specific SKUs. Sean would need to dig into the sheets. Marketing would then want to feature that product more heavily, which meant going back to the sheets again. Information couldn't flow freely between functions.

The result: monthly planning sessions consuming a full day, plus 1-2 hours of daily maintenance to keep formulas from breaking. Classic supply chain pain showed up as stockouts on A-SKUs marketing wanted to push and excess inventory on C-SKUs that weren't moving.

For a low-eight-figure DTC brand where Q4 is 35% of revenue and capital is tied up in inventory, this wasn't just inefficient. It was a strategic constraint limiting how fast they could move.

When Sean moved to start a separate part of the business in early 2024, Aaron (the head of marketing) would take over demand planning. 

The transition created a clean break to question everything. They could either fix the band-aid issues and create a better system internally, or look at what else was out there.

The initial plan was to rebuild in-house. Aaron and the data analyst would design new architecture from scratch. 

Then Sean mentioned Atomic Supply, an AI-powered planning platform he'd heard about at a startup event. They weren't actively looking for vendors, but with this warm lead, they decided to explore what was available.

Most demand planning software they'd seen required heavy transformation: new processes, new philosophies, 4-6 week buildouts plus 2-3 months of onboarding. 

What changed their minds came down to one conversation with Neal. 

Instead of prescribing how Vincero should plan, Neal asked them to explain their current process and promised to figure out a better way to do it alongside their existing system. They wouldn't need to completely abandon what they'd built.

Three things made AI more attractive than rebuilding:

1. Speed to value: Atomic promised a working first draft within two weeks, not months of development time.

2. Flexibility for edge cases: Rather than documenting every exception upfront, they could iterate as real planning cycles revealed gaps.

3. No duplicate data entry: Aaron could keep forecasts in a simple spreadsheet, lead times in another, and Atomic would pull from these single sources.

The decision came down to resource allocation. 

Building in-house meant months of development time with no guarantee it would handle their complexity. AI meant they could evaluate value within weeks and iterate from there.

Atomic's data requirements were simpler than expected. They needed three pieces of information from Vincero:

Neal explains his team's philosophy is to do a lot with a little. 

From there, Vincero layered in the item master (SKU metadata including categories, lead times, and safety stock requirements) and their category-level forecasts. Within two weeks, they had AI-generated recommendations for all 600+ SKUs.

Aaron appreciated this approach because many other tools require you to update your own existing systems and then also update the tool itself. He wanted one source of truth, because anytime you have duplicates, that's when everything starts breaking.

The data architecture stayed simple:

The initial system was roughly 80% there. The remaining gaps would emerge through actual use over the coming months.

Here's the critical decision Vincero made: they didn't switch over immediately. 

They ran their old spreadsheet system and the new AI system side-by-side for six months through their peak season.

Aaron explains the thinking. 

They were going into holiday and had to put their holiday orders in August. Four months into using Atomic, he trusted it as a promising new tool that seemed to be working well. But they'd been using their spreadsheet system for 10 years and it had gotten them this far. They weren't going to completely abandon it.

Every planning cycle, they ran both systems. 

Compared outputs. Investigated discrepancies. Sean notes that the learning curve was more about trust than technical complexity. You're still going to want to keep your existing system in case the new one fails. You can't abandon what you're currently doing.

This doubled their work temporarily. But it eliminated the binary risk. They could validate the AI system while maintaining their safety net through Q4, which represents over a third of annual revenue.

Besides helping them gain confidence with Atomic’s recommendations, the parallel period revealed gaps they hadn't anticipated upfront:

Chinese New Year supply constraints emerged in August, several months after implementation. They realized they didn't just have seasonal demand patterns, they had seasonal supply patterns. The AI needed to factor in supplier shutdowns, not just demand forecasts.

Engraving SKU complexity surfaced as they worked through real orders. Engraving turns the same product into two SKUs (one with engraving, one without). For watches, they also need to buy case backs separately for the engravings. The system needed to track both finished product and component inventory.

SKU classification rules needed refinement. Vincero required different safety stock levels for different tiers:

Exception handling for urgent orders also required iteration. Aaron notes there was an exception to every single rule they had. Their stated lead time was one thing, but in cases where they really needed inventory, they'd force feed orders through faster channels.

"This recommendation doesn't make sense to me because, oh, now I have this new rule that we didn't put into this." - Neal Suidan

This is exactly why the parallel approach worked. 

They discovered these gaps through real planning cycles, incorporated them iteratively, and built confidence that the AI could handle their actual business, not just the simplified version they'd described in week one.

The parallel period did something else: it forced them to articulate why they made decisions. 

When the two systems disagreed, they had to explain which one was right and why. That process built institutional knowledge beyond just trusting one system.

By January 2025 (nine months in) they had enough evidence across different seasons, product mixes, and scenarios. The old sheets are now broken and outdated. They're fully trusting Atomic.

The learning curve for Atomic's AI platform was minimal. 

The overarching process was similar to what they were already doing: reviewing demand, checking supply, looking at POs. But three operational shifts transformed how they work.

Vincero moved from monthly planning that felt like a burden to weekly planning that feels manageable. They weren't looking at inventory weekly before because so much different data was coming in and they didn't have infrastructure to bring it together that frequently.

Weekly cycles enable smaller, more frequent corrections. 

Instead of placing one big monthly order and hoping it's right, they adjust based on what's actually selling. 

Neal sees this pattern across brands.

Aaron and his operations manager now have a meeting every Tuesday afternoon. It's 30 minutes, maybe 45 if it runs over. They start with recommended orders and work from there. Aaron describes it as the AI doing three laps of a four-lap race. Vincero was previously running those three laps themselves.

Before: Full day monthly + 1-2 hours daily maintenance
After: 30-45 minutes weekly

They're spending millions of dollars on inventory but only about four hours a month on inventory planning. 

The time savings enabled more frequent ordering for categories with lower MOQs. When planning took a full day, they batched everything monthly. At 30 minutes per cycle, they can be more surgical about timing.

The first two changes are about speed and efficiency. This third change is about organizational structure, and it might be the most important.

Before AI, Sean (operations) owned the entire planning process. Aaron (marketing) would spot opportunities: a product was trending, a campaign was performing well, they needed to shift inventory priorities. But he couldn't act on any of it directly. He'd have to go through Sean.

"Are we going to be in stock on this SKU?"
"Can we push more inventory to this product?"
"What happens if we feature this in emails?"

Sean would dig into the spreadsheets, run the numbers, and get back to Aaron. 

Sometimes hours later. Sometimes the next day. By then, the marketing opportunity had often moved on. It wasn't Sean's fault. It was structural. Marketing and supply chain were operating in separate worlds, and information couldn't flow fast enough for them to collaborate in real-time.

"Marketing is the speedboat. They want to be able to zip around, move back and forth. From the demand planning side, you're like a cruise ship." - Sean Agatep

Now, Aaron directly owns planning. 

He reviews AI-generated recommendations every Tuesday, makes calls on which SKUs to push based on what marketing campaigns are running, and adjusts forecasts when promotions perform differently than expected. He has the data he needs to make inventory decisions in the same meeting where marketing discusses strategy.

Sean sees it clearly. If they were still using the old sheets, he wouldn't have bandwidth to handle both inventory and marketing responsibilities. The only reason he can wear both hats now is because the process has become simplified and the data is trustworthy.

The organizational shift enabled better SKU portfolio management. 

When marketing has direct visibility into inventory economics, they make different creative decisions. They're more willing to let C-SKUs stock out. They push harder for safety stock on A-SKUs. The conversation shifts from "get me more inventory" to "which inventory actually moves the business?"

Aaron argues this is the biggest ROI: getting marketing closer to final decision-making for inventory planning. 

That was their biggest friction point. Being able to close that gap and have marketing drive those decisions meant they could tighten up how they classify A, B, and C SKUs because marketing now has their hands on the levers.

Most operators evaluate AI tools by comparing features. But there's a simpler way to know if you're ready: look at why you're planning monthly instead of weekly.

Vincero wanted weekly planning. 

They had the business need (seasonal swings, promotional cycles, product launches). But executing felt like too much burden with spreadsheets. So they accepted the lag between market reality and their ability to respond.

That gap between what they wanted to do and what they could actually execute reveals something important about why AI works for this problem.

Before you jump into vendor evaluations, understand why AI is particularly well-suited to this challenge. It's not "AI is the future" hype. There are specific characteristics of planning problems that make them a strong fit for AI.

Complexity scales exponentially, but your capacity doesn't

Think about Vincero's planning challenge: 600+ SKUs across 4 categories with different suppliers, varying lead times, seasonal patterns that differ by product type, and exception rules that change based on urgency. That's thousands of interdependent variables.

You can handle 50 SKUs with spreadsheet formulas. Maybe even 100. But at 600+, every new rule you add creates exponential complexity. The formulas become brittle. One wrong adjustment breaks the entire system.

AI handles this differently. It doesn't use rigid formulas. It learns patterns across all your variables simultaneously and adapts when you add new rules.

Pattern recognition you can't do manually

AI spots correlations you'd never find in spreadsheets. Which products move together? How do promotional patterns differ between categories? What seasonal patterns emerge when you look across multiple years of data?

For Vincero, this meant understanding that their seasonal patterns weren't just about demand (Q4 surge, Valentine's Day spike, Father's Day bump). They also had seasonal supply constraints around Chinese New Year. AI could factor both into recommendations in ways spreadsheet formulas couldn't.

Speed enables a fundamentally different planning cadence

The real unlock isn't that AI makes planning faster. It's that AI makes planning so much faster that you can change your entire approach.

When regenerating a plan takes seconds instead of hours, you move from monthly planning to weekly planning. That shift matters more than the time savings. Sean's insight captures this: 100% of forecasts are wrong, so the point isn't perfect predictions. It's frequent corrections.

Weekly planning means smaller adjustments based on recent data, not big monthly bets based on month-old assumptions.

Scenario planning becomes instantaneous

"What happens if we drop this SKU by 20% and push that budget into the bestseller?"

This was the question Aaron kept asking Sean. The question that would take hours to answer because Sean needed to manually adjust formulas across multiple sheets.

With AI, you get the answer in seconds. That changes what kinds of questions you can ask. When scenario planning is instant, marketing can actually collaborate with supply chain in real-time instead of waiting days for answers.

The system learns from what actually happens

Spreadsheets use static formulas. You predict demand will be X, lead time will be Y, and you place orders accordingly. When reality differs from your predictions, you manually update your assumptions for next time.

AI continuously learns from what happened versus what was predicted. It sees that your A-SKUs consistently sell 15% above forecast during promotional periods. It notices that certain suppliers are reliably 2 weeks behind their stated lead times. It factors this into future recommendations automatically.

Exception handling without breaking the system

This might be the most important point. Real businesses have exceptions everywhere.

"Lead time is 8 weeks, unless we need it urgently and force feed it through in 4 weeks." "Safety stock is 2 weeks, except for A-SKUs which need 3 weeks minimum." "Limited editions should come out of the category forecast but use different reorder logic."

Spreadsheets require perfect logic trees for every exception. Add too many and the system becomes unmaintainable. Aaron described their old system as having an exception to every single rule, which is why it kept breaking.

AI handles exceptions differently. It learns patterns rather than following rigid rules. It can understand "usually do X, but in these circumstances do Y" without you having to encode every possibility upfront.

These advantages matter most when you're dealing with specific types of complexity. Here's how to assess if AI planning makes sense for your business:

Vincero's experience gives you a roadmap. Not a perfect blueprint, but a set of decisions you can adapt to your situation. 

Here's what matters if you're evaluating AI planning tools.

When someone on your team leaves, gets promoted, or hands off planning responsibilities, that's your moment. You're not defending past decisions anymore. You're figuring out what works going forward. Don't wait for a crisis. Use these natural transition points to ask if there's a better way to do this.

Vincero got an 80% solution in two weeks. Traditional tools promise you completeness after 4-6 week buildouts plus 2-3 months of onboarding. Ask your vendors: can you show me something working in two weeks, even if it's incomplete? Their answer tells you everything about their approach.

You won't catch all the edge cases in scoping sessions. They'll emerge when you're running actual planning cycles. Chinese New Year supply constraints. Engraving SKU complexity. Exception handling for urgent orders. Vincero discovered all of these after they started.

So test your vendor's flexibility early. Give them a complex edge case from your business. Watch how they respond. Do they want to scope it fully before starting? Or are they comfortable saying "let's get the basics right and iterate when we hit that"? The second approach works better when your business rules keep changing.

Before you talk to any vendor, get your data house in order. You need three things in queryable form: sales history, purchase order history, and inventory on hand. If you're maintaining duplicates across multiple systems, fix that first. Aaron's point is simple: anytime you have duplicates, that's when everything starts breaking.

Walk through your current data architecture and ask:

If you don't fix these issues first, you're just moving a mess from spreadsheets to software.

This is non-negotiable. You're making million-dollar inventory decisions. Run your old system and the new AI system side-by-side for at least one full seasonal cycle. Yes, it doubles your work temporarily. But it eliminates the binary risk.

Make sure your parallel period covers:

The parallel period does two things. First, it validates the AI can handle your actual business. Second, it forces you to articulate why you make decisions. When the two systems disagree, you have to explain which one is right and why. That builds institutional knowledge you can't get any other way.

Even with AI recommendations, you need to apply judgment. Aaron checks specific SKUs every Tuesday because he's on the marketing side. He has context the algorithm doesn't fully capture yet. Your promotions. Your seasonal campaigns. Your gut feel on which products are about to take off.

AI should accelerate your judgment, not replace it. If a vendor tells you their system "makes decisions for you," walk away.

Time savings are easy to measure. Vincero went from a full day monthly to 30 minutes weekly. That's a 90% reduction. But the bigger unlock was organizational.

Ask yourself these questions:

Those answers show you where the real opportunity is. For Vincero, it was bringing marketing into planning decisions. That changed how they manage their entire SKU portfolio. Your unlock might be different, but you won't find it by only looking at operational efficiency.

Vincero's experience revealed insights that changed their entire approach to planning. These weren't things they expected. They emerged through actually using AI for daily decisions, and they'll probably shift how you think about your own process.

Sean's insight matters: 100% of forecasts are wrong. So stop trying to perfect your monthly forecast. Start thinking about how quickly you can correct when reality diverges from your plan.

Weekly check-ins mean you're adjusting based on recent data, not month-old assumptions. This shift from "plan perfectly" to "plan frequently" will change how you allocate your time. You'll spend less energy trying to nail the forecast and more energy responding to what's actually happening in your business.

You're probably evaluating AI tools for time savings. That's fine, you'll get them. Vincero cut planning time by 90%. But here's what surprised them: the organizational alignment mattered more.

When Aaron (marketing) got direct access to inventory planning, everything changed. Marketing made different creative decisions. They pushed harder for safety stock on A-SKUs. They were more willing to let C-SKUs stock out. The conversation shifted from "get me more inventory" to "which inventory actually moves the business?"

That organizational unlock will probably be your biggest ROI too. You just won't see it in a spreadsheet comparing time before and after.

Vincero is still finding edge cases nine months in. Preorders. New supplier constraints. Shifting promotional calendars. There will always be something that needs adjustment.

Here's the shift in thinking: the advantage isn't having a perfect system. It's having one that can evolve as your business evolves without rebuilding from scratch. When Aaron says they're still figuring things out and that's fine, he's describing a different relationship with planning tools. You're not looking for something that handles everything perfectly on day one. You're looking for something that can adapt quickly when your business changes.

That flexibility matters more than getting everything right upfront.

So here's where this leaves you. 

The question isn't whether AI can improve your planning. It's whether you can afford to keep running monthly cycles when your business needs weekly decision-making.

For Vincero, AI-powered planning meant cutting time spent by 90%, moving from monthly to weekly cycles, and bringing marketing directly into supply chain decisions. 

The transition took nine months of parallel testing and iterative refinement. But the result was a planning process that could finally keep pace with their business.

The parallel system approach they used wasn't just smart risk management. 

It was what made the entire transition possible while running a real business through peak season. If you're facing similar constraints, struggling with the same spreadsheet complexity, dealing with the same organizational friction between marketing and supply chain, their approach gives you a way forward.

You don't have to blow up your current system to improve it. You just need to run both until you've built enough evidence to trust the new one. That's how you get from where you are to where you need to be without risking the business you've already built.