Is it worth doing something that you know is going to fail?

What I Learned at the Intersection of Restaurant Ops and AI

In mid-March, Fourtop wrapped its first Restaurant AI Assembly working group. Facilitating and supporting a vibrant community around the use of data is a foundational pillar for Fourtop, and this project – an effort to “explore the profitable, sustainable & human application of automated decisioning in the hospitality industry” – is one to which the leadership team has been committed since my first day. The Restaurant AI Assembly brings together experts with varied experiences from within and outside the restaurant world to explore new ways to use the automation of data to address common and complex problems. The goal is to provide actionable guidance to restaurants and operators on the use of data and AI to improve performance, and since I have been finishing up my master’s degree in system-based training, I was asked to contribute to the Assembly’s final outputs by providing business use case studies, SOPs and training materials that would help demonstrate how new AI models and processes could be adopted by restaurants.

The two working group sessions had gone extraordinarily well, with participants bringing a wide range of insights and suggestions to the table. However, while the completion of a successful first working group should have brought a sense of accomplishment, instead it felt like we had just opened up a whole new can of worms.

The Restaurant AI Assembly Project

Projects always feel the most achievable before they have begun. The first question we identified and decided to explore was how to improve the efficacy of labor management models and systems for restaurants. As a former director of ops for a multi-unit full service restaurant group, and someone with more than a decade in restaurant operations, I can describe in detail the many issues involved with managing employees, and also the many ways in which the tech systems available to operators can fail us. I felt confident that between the innovative strategy and data team at Fourtop, and some of the most accomplished experts on the subject of labor management and employee relations, we would identify a brilliant path forward. However, upon the conclusion of the second session, I was convinced we were biting off more than we could chew.

In the first session we presented an initial theory around labor management to the participants, and it was torn apart in a matter of minutes—surprisingly, that felt great. We then spent the entire session discussing how we can’t even begin to solve a problem when we don’t have a clear idea on the intended outcome. The concept had assumed that an automated labor management model would be oriented around optimizing the deployment of labor for efficiency, while managing employees through a cycle of onboarding, training, performance evaluation, and an eventual exit. By the end of the session it had become clear that every part of our initial model, from the basics of an optimized schedule to a method for employee performance tracking, would need to be reimagined. We had started the session with something, but we ended it with nothing, except infinite possibilities and too many ways that we could do this badly. It was beginning to feel like even a basic workable labor management model was going to be an impossible task.

In the second group session, we presented a concept for a holistic, data-driven employee relationship management model, which would seek to derive more value from employee performance through smarter, more adaptive employee engagement. Again, as we worked through the approach with the group, it was clear that there were so many details to consider, we were just scratching the surface.

I finally said out loud what I had been silently asking myself for the last month: why did we think we could solve this problem? Nearly every restaurant operator deals with labor management challenges, yet there are hardly any solutions on the market that really improve the experience of both the employee and the manager while driving positive business outcomes. How can this one group possibly make a difference? And why am I being asked to contribute? Me. The person who googles every acronym during calls because I’m still learning the jargon of a data technology company.

My first job in my hometown's local bake shop (circa late 2000s)

Why I hope I can help

My background, up until joining Fourtop, has been in restaurant ops. I was the 17-year-old taking my high school’s Intro to Culinary Arts class and getting all of my friends jobs at the local bakery on weekends and after track practice. I pursued my undergraduate degree in hospitality administration and I fell hard for the excitement of operating an all-in-one business like a restaurant. I went to work in full-service restaurants after graduation and came to understand that restaurant operators are jacks-of-all-trades. I learned how to fix the plumbing, write a schedule, rewire a lamp, maintain inventory par levels, clean a fryer, bartend, and grill a pretty mean cheeseburger. I loved learning all of these things and I loved teaching others how to do them as I moved from General Manager into Training and Development and then into a Director of Ops role.

So why did I leave Ops?

Going to work at a tech startup, with intelligent, restaurant-focused, stand-up individuals, was never a part of any of my 40-year career plans (yes, I have a few). I want to say it's because I would like to impact more operators than I can at an individual restaurant company, and there is a lot of truth in that. Given my studies in instructional design and technology, I wanted the opportunity to help the industry be more efficient.

But also, I was tired. After being an end user, and then eventually a buyer, of restaurant technology for the last 10 years, I was tired of having to find workarounds and ways to just get the tech we employed to function how it was promised to function. I wanted to be a part of helping technology companies figure out how to fill an operator's cup, not drain it.

To be clear, I think the restaurant industry is better because of the tech we employ, but it’s not perfect— and the gaps between the way many technology systems function and how restaurant operations need them to function can cause a lot of pain. As an operator, and particularly a manager, my life had become a constant blur of functional silos, data re-entry, and manual corrections and hacks.

I wanted to help make things better, and yet here I was on my first project, working on a problem that was directly related to the work I’d been doing for a decade, and I felt like we were going to fail, almost before we had started.

Are we biting off more than we can chew? (Spoiler: 100% yes.)

In the last six months since making the transition to the world of restaurant tech, I’ve come to learn that a technology company’s goal is to find the simplest, most widely applicable path to solving a common problem. Its binary–black and white. But restaurant operations are full of complexities that don’t magically conform to this one-size-fits-most requirement. Even the best operators with the tightest SOPs know that procedures and policies have a time and place to be enforced. We wouldn't terminate a high performing, team player who had car trouble and deserves some grace. And we can’t just tell a guest with two hungry kids to come back in three minutes when the doors “officially” open for the day. That’s not what hospitality is.

Tech can’t just simplify human complexities with the flick of a switch—at least not quickly or easily. I have seen how Fourtop’s technology can bring data together at scale from many systems with both precision and accuracy to enable new ways of using it, but I also have seen how many details are needed for even one new data model to work for a single restaurant, let alone the entire industry. Solving common problems at scale, even with advanced data technology, isn’t a first shot thing. My experience facilitating this first session of the Restaurant AI Assembly was teaching me that, in technology, making progress is incremental…slow…a constant act of failure.

The findings and outputs from this first working group, which Fourtop will be releasing next month, don’t feel like enough. And we know that. We want to help inform how the industry tackles an incredibly complex problem, and to provide direction on how data, technology and AI can not only make the business of restaurants more profitable, but also the work within restaurants better for employees. There is so much to be done, and Fourtop is only one piece of the puzzle.

It is obvious to me Fourtop is committed to this journey. The team is dedicated to testing theories and garnering feedback from the industry. They (well, we) are also committed to grounding theory in the real needs of operators and to presenting actionable guidance to use their data. And while the work always will be ongoing, perhaps never complete, it doesn’t deter from striving for the ideal.

My future as an operator

I was recently struck by this question: am I still a restaurant operator?

It’s been a core part of my identity for the past decade. And I take a lot of pride in the designation.

Good operators can dive down into a department or a line item on a P+L and then look at the numbers holistically to see the bigger picture. We are flexible and adaptable. Most importantly, we are responsive to our people, whether they be our employees, guests, or our partners.

Entering the tech world I initially felt as if everything was different— that there was a whole new set of issues and completely different stakes to these problems. But I’ve come to realize that, while some of the trappings may be different, and while I haven’t had to write a prep list or floor manage a crazy Sunday brunch recently (we all know Mother’s Day is right around the corner!), the work is still about trying, and occasionally failing, to make restaurants a little more hospitable.

I may not be always working within the four walls of a commercial kitchen anymore (though I do find myself there a lot these days working with our clients), I think that the operator mentality of always making things work and always delivering for the guest will forever remain with me. And from here on out, always recognizing that failure is the best next step on the path to success.