Jersey Mike’s IPO illustrates how bad the AI hype has become

Jersey Mike's IPO illustrates how bad the AI hype has become

When a submarine sandwich chain files for an initial public offering and name-checks artificial intelligence 22 times in its S-1 registration statement, the AI hype cycle has officially jumped the shark — or in this case, the sub.

Jersey Mike's, the fast-casual franchise known for sliced-to-order cold cuts and Danny DeVito commercials, isn't building large language models. It's not selling predictive analytics dashboards. It makes sandwiches. Yet its investor prospectus reads like a Series A pitch deck from 2023, sprinkling "AI Technologies" throughout risk factors and strategy sections with the same fervor as a venture-backed SaaS startup chasing a term sheet.

The AI-washing playbook has gone mainstream

We've watched this pattern unfold across the CRM and sales technology landscape for two years. Vendors who couldn't spell "transformer architecture" in 2022 suddenly launched "AI-native" rebrands. Legacy platforms bolted on generative features — often thin wrappers around OpenAI APIs — and called it innovation. Buyers learned to ask sharper questions: Where does the model run? What data trains it? What's the hallucination rate? What's the actual ROI?

Now the performative signaling has migrated to main street. Jersey Mike's discloses it is "beginning to use AI Technologies in our business" — a phrase so vague it could mean anything from a ChatGPT prompt for social media captions to a demand forecasting model for turkey orders. The filing mentions software 52 times and data 112 times, which is honest: modern franchise operations run on POS platforms, inventory systems, and CRM tools for franchisee management. But the AI risk disclosure reads like boilerplate copied from a tech company's template, not a sandwich operator's material exposure.

Real AI risk vs. performative disclosure

To be fair, the restaurant sector has genuine AI cautionary tales. Starbucks recently scrapped its Deep Brew inventory management tool after it failed to accurately predict demand — a reminder that bad models produce bad operational decisions at scale. But Jersey Mike's S-1 mentions weather-related risks five times and lightning strikes zero times, despite a documented 2021 franchise location lightning strike in Texas. If statistical probability guides disclosure priorities, AI catastrophe ranks well below acts of God.

This matters for CRM and RevOps leaders because it distorts buyer expectations. When every vendor — and now every sandwich shop — claims AI transformation, the signal-to-noise ratio for evaluating actual capability collapses. Procurement teams waste cycles separating substance from marketing. Budget holders approve tools based on hype criteria rather than workflow fit. The entire category suffers credibility erosion.

What buyers should actually ask

The Jersey Mike's filing is a useful litmus test. If a sandwich chain can stuff 22 AI references into an S-1 without shipping a single model, then any CRM vendor's "AI-powered" claim deserves the same skepticism. Smart buyers are shifting evaluation criteria:

• Architecture transparency: Is the AI a proprietary model, fine-tuned open source, or an API call to a third party? The answer determines data governance, latency, and cost structure.
• Workflow integration: Does the feature eliminate steps or add a chat sidebar nobody uses? Jersey Mike's doesn't need a generative sub-builder; it needs accurate labor forecasting. Your sales team doesn't need a poem generator; it needs pipeline risk scoring that works.
• Measurable outcomes: Vendors should cite customer benchmarks — conversion lift, cycle time reduction, forecast accuracy improvement — not just "AI-enabled" badges.
• Failure modes: Ask what happens when the model is wrong. Starbucks learned the hard way. Your CRM forecast roll-up has the same downside.

The hype hangover is coming

Markets eventually punish performative innovation. The 2024-2025 CRM buying cycle already shows fatigue: shorter proof-of-concept windows, stricter procurement reviews, and a revival of "boring" but reliable workflow automation over flashy demos. Vendors who invested in real ML infrastructure — think Salesforce Einstein GPT's trust layer, HubSpot's Breeze intelligence, or Gong's conversation analytics — will separate from those who shipped wrappers.

Jersey Mike's will likely go public, sell subs, and never deploy a single transformative AI model. Its investors will be fine. But the episode signals a peak: when the most analog business in the filing cabinet feels compelled to cosplay as an AI company, the bubble isn't just inflated — it's begging for a pin.

For CRM buyers, the lesson is simple. Ignore the AI count in the pitch deck. Count the problems it solves.