Give an Ovation: The Restaurant Guest Experience Podcast with Zack Oates · Ovation

Making AI Practical in Restaurants with Carl Orsbourn of Invisible

January 20, 2026·22 min·1 clip
Carl Orsbourn reveals that 95% of generative AI pilots fail to scale, according to recent Gartner data.
1. Zach Oates hosts Carl Orsbourn, SVP at Invisible, on Give An Ovation to discuss practical AI adoption in restaurant and hospitality businesses. 2. Carl Orsbourn is a multi-time podcast guest, best-selling author, and SVP at Invisible, a 10-year-old company that did approximately $130 million in revenue last year and was ranked the second fastest-growing AI company in the US. 3. The episode's core thesis is that AI adoption fails when framed as cost reduction and succeeds when framed as a force multiplier that frees humans to do what only humans can do. 4. Invisible was involved in training the first iteration of ChatGPT and continues to support large language model training, working with companies including OpenAI. 5. Gartner data cited by Orsbourn states that 95% of all generative AI pilots launched in the past 12 months failed to scale, typically because they lacked clear measurement, shaky data foundations, or insufficient organizational champions. 6. Invisible's platform has three modules: a data structuring layer using a medallion architecture, Atomic (a custom process builder that works alongside existing tech stacks rather than replacing CRMs or ERPs), and Axon (agentic workflows). 7. The Radial marketplace module gives Invisible access to 20,000 people globally who assist with model training and human-in-the-loop quality assurance, which Orsbourn says any vendor claiming 'AI does it all without humans' is misleading. 8. Orsbourn introduces the '10 fewer people vs. 1,000 more people' reframe: a dairy farmer who thought of AI as headcount reduction would never have arrived at the computer vision drone system that now monitors cattle welfare and detects early signs of infection. 9. The drone system frees the farmer's human workers to respond faster to individual cow health issues, improving yield — illustrating how the question 'what would you do with a thousand more people' unlocks use cases the deficit framing misses. 10. Orsbourn says the restaurant industry parallel is clear: the best GMs spend the most time with their team and guests, yet most GMs spend the majority of their time on computer tasks and administrative work. 11. He proposes a 'Smart GM' AI assistant concept that would simultaneously handle staff call-outs, predictive equipment failure alerts, demand forecasting for weather-driven delivery surges, and cross-location part-time staff redeployment. 12. The Smart GM example involves two cooks calling out, a refrigerator sensor predicting failure within a week, and a snowstorm arriving the same evening — the AI would coordinate technician booking, cross-staff deployment, and expo station reassignment without GM input. 13. Orsbourn cites five specific data points for guests who receive proper service recovery after a bad experience: 24 times more valuable, twice as likely to return, $5 more per subsequent visit, 2.5 times average visit frequency, and 12 times more likely to leave a five-star review. 14. He frames AI's role in hospitality through the Tony Stark Iron Man suit analogy: the technology amplifies human capability rather than replacing it, and the humans who make eye contact and say 'I'm glad you came in' are irreplaceable. 15. A major near-term obstacle Orsbourn identifies is disconnected SaaS stacks: a restaurant executive may have separate demand forecasting models operating in their labor tool, inventory tool, business development team, and finance team simultaneously. 16. He argues that bad data defeats good AI every time and that fixing the data infrastructure is a prerequisite for AI to deliver on its potential in hospitality. 17. The conversation's tone is enthusiastic and forward-looking, with Oates matching Orsbourn's energy and both finding analogies in switchboard operators, the dishwasher, the Segway, and Tony Stark. 18. The episode is structured as a wide-ranging framework discussion rather than a tactical how-to, staying mostly at the strategic and conceptual level. 19. Restaurant technology executives, operators thinking about AI strategy, and hospitality investors weighing AI vendor claims will find the most value here. 20. Operators looking for specific vendor recommendations, step-by-step implementation guides, or restaurant-specific case studies rather than frameworks will need to look elsewhere.

As heard by us

A practical AI discussion that keeps returning to restaurant operations and guest experience.

Carl Orsborn returns for a third turn on Give and Ovation, and the conversation stays close to the realities of restaurant operations. Zach Oates keeps the attention on staffing, service, and the practical upside of more capacity when the aim is a better guest experience and…

Read the full review in PlayNext →

Why you'd press play

When you want AI advice that sounds useful on the floor, not at a keynote.

Read the full recommendation in PlayNext →
Listen to the show on