David vs. Goliath: AI, Excel, and the Work No One Talks About

By: Vaishu Myadam 

Every week, a new AI breakthrough captures attention. ChatGPT enters healthcare. DeepSeek releases cinematic video models. Autonomous agents begin scheduling meetings, writing code, and negotiating contracts. Claude recently released financial agents capable of working within Excel and PowerPoint. The future of work feels increasingly futuristic—fast, visual, and intelligent. 

Yet the most economically meaningful AI revolution may not be happening in hospitals or Hollywood. It may be happening inside spreadsheets. 

Across millions of small and medium businesses, procurement—i.e. deciding what to buy, when to buy it, and how much—still runs on fragile Excel formulas, email threads, and tribal knowledge. It is invisible, operational, and deeply inefficient. It is also one of the largest untapped opportunities for AI-driven change. 

At Kellogg, a student-founded startup, Nyck AI, is building an “AI employee” for procurement—not to create flashy demos, but to automate the kind of repetitive, cognitive labor that quietly drains time, money, and attention from growing businesses. 

From left: founder Dániel Sütheő, founder Marc Davis, CTO Deepika Sampangi Credit: Shane Collins

Founded by Marc Davis (MMM ‘26), Dániel Sütheő (2Y ‘26), Alexandre Rossi Alvares (2Y ’25) and Arthur Koefender (2Y ’25), Nyck AI recently became a semifinalist for the Microsoft Imagine Cup, a student competition that offers aspiring founders an environment to build, validate, and accelerate startups with the support of Microsoft’s AI technologies and global ecosystem.

Their work reveals something bigger than a single venture: the next wave of AI won’t glamorize work. It will erase it. 

When Automation Becomes Labor 

Enterprise software has promised efficiency for decades. ERPs, CRMs, and procurement platforms digitized workflows but rarely removed them. Humans still plan purchases, reconcile spreadsheets, generate orders, and manage exceptions. 

Nyck AI’s approach reframes this model. Instead of offering another dashboard, the team built an AI agent that learns purchasing behavior, forecasts demand, calculates optimal order quantities, and generates purchase orders. Humans stay in the loop for oversight, but the operational burden shifts to software. 

This distinction matters. Traditional software improves workflows. AI agents replace them. 

In this framing, businesses aren’t buying tools—they’re hiring labor. Procurement becomes less about managing systems and more about supervising autonomous decision-makers. This shift aligns with a broader trend in AI: from software as infrastructure to software as workforce.  

As multimodal models advance, digital labor markets are emerging, where intelligence becomes something organizations can rent on demand. 

Procurement, despite being deeply unglamorous, becomes an ideal proving ground. 

Why the Real Opportunity Lives in Small Businesses 

Nyck AI initially assumed large enterprises would be the primary customers. After months of interviews, they discovered the opposite. The deepest pain lived in small and mid-sized companies.

Unlike corporations with sprawling ERP stacks, many distributors and manufacturers still operate on spreadsheets patched together with email. For firms managing thousands of SKUs, even minor inefficiencies cascade into stockouts, bloated inventory, and operational stress. 

For these businesses, AI isn’t about innovation—it’s about survival. Small firms are more agile, more resource-constrained, and far more sensitive to efficiency gains. Saving hours of labor or improving forecasting accuracy by a few percentage points can materially change margins and cash flow. 

This reflects a larger shift in AI adoption: transformation is happening not at the elite enterprise tier, but across the operational long tail. 

The Platform Risk: Can Giants Simply Absorb This? 

There is, however, a powerful counterargument: Can’t incumbents simply build this? 

Salesforce, SAP, Oracle, and Microsoft already sit at the center of enterprise workflows. Apple routinely absorbs entire startup categories by shipping native features. If procurement agents prove valuable, platform giants could surely integrate similar functionality, compressing entire startup markets overnight. 

This risk is real. History suggests that defensibility in software is fragile. The founders of Nyck AI acknowledge this tension, and their response is twofold. 

First, large platforms typically serve large enterprises. Their tools are expensive, complex, and slow to adapt. Meanwhile, Nyck AI targets the millions of smaller businesses priced out of enterprise-grade software—firms that need speed, simplicity, and affordability more than deep customization. 

Second, speed of innovation matters. Big companies move deliberately. Startups move urgently. The gap between “possible” and “shipping” creates space for experimentation, iteration, and early customer loyalty. 

And even when platforms do absorb features, they often validate entire categories. Apple didn’t kill mobile startups—it created them. Many were acquired. Others thrived by serving niche audiences better than platform defaults ever could. 

In this framing, acquisition is not failure. It is one of the natural endpoints of successful platform-layer innovation. 

Credit: https://www.nyck.ai/

|The Human Layer Still Matters 

Despite ambitious claims about autonomy, Nyck AI’s system is intentionally designed with human oversight. Procurement is riddled with nuance: volatile demand, supplier disruptions, and business-specific constraints that defy rigid logic. Rather than eliminating humans, the system acts as a first-pass decision-maker—generating recommendations and purchase orders that people validate. This hybrid model reflects a broader truth about AI deployment: trust, not capability, is the limiting factor. 

In high-stakes domains—healthcare, law, finance, and operations—humans remain accountable. AI shifts the burden of execution, allowing people to focus on judgment, strategy, and exception-handling. 

The future of work may not remove humans. It may elevate them. 

Kellogg as a Venture Discovery Engine 

 

Nyck AI’s evolution highlights Kellogg’s distinctive role in shaping AI entrepreneurship. The company’s earliest customers emerged directly from student interviews. The campus functioned as a live experimentation lab surfacing real operational pain rather than abstract startup ideas. 

Instead of beginning with a fixed product vision, the team started with a problem space. Through dozens of conversations, they pivoted from enterprise procurement to SMB operations, following lived business reality rather than theoretical market size. 

In an era where AI tools evolve weekly, this approach offers a durable advantage. Technologies change quickly. Human problems do not. 

From Software to Organizational Design 

 

Nyck AI’s framing of its product as an “AI employee” hints at a deeper structural shift. As companies increasingly hire intelligence instead of labor, organizational design itself will change. 

Teams may shrink. Roles may become more strategic. Operational layers may compress. 

Businesses will not simply adopt new tools—they will redesign how work happens. 

The next AI revolution won’t arrive with spectacle or headlines.  It will slip quietly into spreadsheets, inboxes, and workflows, dissolving tasks so mundane we barely realized they were work at all. 

And when we finally notice the change, it won’t be because of what AI has created, but because of what it has made unnecessary. Read More Tech Talks: Everyone Says 'Just Build.' I'm Not So Sure. Why the Builder of an AI Editing Service Says He Isn’t Replacing Editors Data, Deals, and Government. What’s New with Nuclear? The Wall-less Empire: The Illusion of Open AI Markets The SpaceX Diaspora: How Rocket Science Is Transforming Startup Culture