AI Making Restaurants Smarter And Safer With Video Intelligence (Ep 181)
The power of video has come a long way! In this episode, Jaime Oikle of RunningRestaurants.com talks with Rish Gupta, Co-founder and Head of Product at Spot AI, about their AI camera system designed to make restaurants safer and smarter.
It was fascinating to learn everything the system can do—from providing visibility across multiple locations to capturing "training moments" and spotting waste and theft to managing incidents and accidents. There are lots of other cool features as well. It truly is "the easy button for video footage."
Be sure to check out the episode!
Find out more at Spot AI and Running Restaurants.
Thanks to our episode sponsor, Popmenu.
---
Watch the episode here
Listen to the podcast here
AI Making Restaurants Smarter And Safer With Video Intelligence With Rish Gupta
In this episode, I speak with Rish Gupta, the CoFounder and Head of Product at Spot AI. We talked about how the easy button for video footage helps restaurants. Stay tuned.
---
Welcome to the show. We're bringing the tips, tools, and techniques you need to know to make your restaurant more profitable and successful. We've got a great episode for you with Rish Gupta, the CoFounder and Head of Product at Spot AI. Welcome, Rish. How are you?
Good. Thanks for having me. I’m super excited to talk about it.
This is tech stuff. I love talking about the tech and we met briefly before and we were talking about a few things, but let's get into the introduction of what you guys do and then I'll dig in from there.
What Spot AI Does
What we do is at Spot AI, we provide a video intelligence platform for businesses to have safer, secure, and better operations. What that translates into in a very simple way is if you have security cameras or cameras in general in your restaurants, our system can connect with any and all cameras, and we have a lot of AI built-in, so you can get reports on the number of people, the heat maps some. Is somebody present behind the counter? A bunch of such examples include incident resolutions. If you have something happened, you want to record that video, use it for collaboration or right on top of those videos. We make your cameras supercharged and work for you.
One of the things I saw on your site that grabbed my attention was that cameras are dumb. They've used to be, but now they're smart and you can use them in restaurants along with a lot of other folks that aren't using them to their full potential. I'm assuming that's where you guys plugged in and said, “This stuff's not being used. The data's out there. Let's find a way for the computers to get into it.” What was the derivation? Where did you guys get the origination from?
We were talking to customers, and what's happening in the technology world is that the cost of cameras is getting cheaper because every mobile phone has a camera. In economies of scale, people have figured out how to do it cheaply and at scale. The chips, which are going to make these cameras smarter, are getting cheaper and smarter and better every year. We looked at that and were trying to build new cameras, but when we spoke to our customers, most said they already had security cameras. Unfortunately, these security cameras were built by hardware companies with relatively poor software.
We said, “We don't need to sell the lenses. We don't need to sell the cameras. If we can power software, the video files will come. Can we make those video files more useful? Can we make them more impactful?” That's exactly what we went and pitched to our first few customers. We were discussing one of them, a restaurant chain based in California and Nevada, and they were like, “This makes sense.” They had 14 or 15 locations then they have 20 locations now. They started with 1 or 2, and then I think we are now across all the locations.
Primary Uses Of Video Intelligence
I could see it for restaurants. What are some of the biggest uses? The first thing that comes to mind, obviously, you see the camera in the sky, Big Brother monitoring theft. Is that one of the bigger ones? Incident stuff? What else?
We work best when there's a multiple-restaurant chain. The owner-operator has to manage either multiple franchisees or multiple restaurants. It gives them a single pane of glass to look at all the cameras. You don't have to switch between different video recorders. You can mix and match cameras from different locations and have a view on your cell phone or on your desktop and see what's happening behind the counter. If you want to know it's 1:00 PM, how's the lunchtime traffic doing it at the next restaurant, you can pull it up. For owner-operators, the visibility across multiple locations without having to make phone calls, without having to wait for some report to come at the end of the day, getting the full visual sense of what's happening in the restaurant, is incredibly important.
Real-time visibility across multiple restaurant locations without the need for phone calls or end-of-day reports - that's the power of advanced video intelligence platforms.
Every once in a while there is an incident that happens, whether it's to do with a customer complaint or an employee complaint. You can then get the customer saying, “The server was rude to me. The person behind the counter did not get their order wrong and that led to a fight or altercation.” They can pull the video up. They're able to investigate for themselves what AC happened, and very often they might realize the person be their staff is not in wrong or it's a training moment for their staff.
Rather than being Big Brother, we promote adding all your people to this platform. Even your employees and staff can see which cameras are there. They can see what's happening behind the counter. It can lead to very interesting results. One of the early stories I remember, which was interesting, was when they saw that the ingredient consumption at two of the restaurants in the chain was much higher than the rest. They didn't realize why it was happening, and then through the videos, they were able to see that in certain sandwiches, some of their staff were loading it up.
They're like, “This is why our cost at these two locations are better,” then they're like, “If you do a training for these few employees and then they can figure out by watching videos which ones are doing it.” They're like, “We need to train them saying, ‘The other thing you're doing wrong.’” They're able to bring it down.
We've all been, whether it's a sub shop or another place and there's the person who does the scoop, levels it off and puts it on, then there's the one that does the scoop and heaps it on, does 3 of those instead of 2. You can definitely see that it is a training moment for sure. I'm seeing some interesting use cases, and a couple of questions have already come to my mind. I want to go back to the heat map for a second and the idea of seeing customers in your location. That is a very useful thing. People always want to know the headcount. You used the phrase heat map. What do you mean there?
There are two things we do with respect to heat maps. 1) Even without doing any AI setup on your camera, you don't have to set up anything. You don't have to toggle anything on. We give a people count throughout the day so you can on any camera. If your camera is looking at the main floor, you can look at, “At 1:00 PM, there were 27 people. At 2:00 PM, there are 25 people.” We give you people count across the time of the day for the last seven days without you having to set up anything. If you wanted to take it a step further and say, “Where were these 27 people? Do people prefer to sit in certain spots off the restaurant? Are certain tables always empty?”
How much traffic are you getting in certain sections? Sometimes, you will have multiple food stations until the way where your ketchups are and stuff like that. One of them is utilized a lot more than the other, like 90% of the people use this one and 5% because of the way it's situated across the restaurant. You are able to see these patents. You're able to see where the human traffic is moving. It is not just the count, but you get a sense of how the traffic is moving within your restaurant. Are there certain sections of your restaurant that are highly underutilized?
You might be surprised with some of those findings. What about the coverage? You talked about it already. People should know where the cameras are and where they're pointing at. Sometimes in a system, people will know where the so-called blind spots are, but what percentage of coverage of ground do you need or want? What do you think?
I think what we at least tell our customers is there is a service area and the staff can have a private office or stuff like that. The service area where the customers are and where the food is getting prepared are areas where you need monitoring because that's the heart of your business, and that's what you're trying to improve. That's what you're trying to get information on. If the employees have a private room in the back where they can go and take a break, chill, put their stuff, maybe that's not where you need cameras because it doesn't add to your operations and you want people to be able to be in a room without having a camera in the room. That's typical advice and different owners have different preferences, and so forth.
One of the notes that we can come back to is and you mentioned it before, accidents happen, slip, fall, and so forth. What's the process? It is probably pretty easy to now you can identify in the AI, the day part. You can go back, you can share the files. What does that look like?
Typically, when an accident happens, and this was surprising to us when we got into the industry, people are using traditional video recorders. There is basically a video recorder on site, which is recording from all the cameras, and you have a fast-forward or rewind button. Let’s say you have to download something from this site to be able to send to a legal department, or you want to send it to a few employees to discuss a safety procedure or something, which is a better practice. You need to literally take a USB thumb drive, take that video out, download it on Dropbox, create a Dropbox link, and then send it on an email and then chat through it.
Simplifying Incident Management
What you can do on our platform is single click, add it to a case, and then you can add all your employees who you are relevant to this case to that case by adding them because they're in the system by adding their name. As the video is playing, you can literally write on top of the video saying, “At 2:30, you see this person overloading the sandwich.”
Streamline incident resolution in your restaurant with seamless video collaboration. Tag employees, add comments, and review multiple angles – all in one place.
Let's say the manager's name is Mark, and this person is Joe. It's like, “Mark, it seems like Joe needs some help with what the right quantitative ingredients need to go on sandwiches. Can you help him train on that? Let's review this again next week.” When Mark and Joe come in into this video, they know exactly which point of the video you're talking about. Automatically, Mark and Joe are tagged to this comment, like you would do it on Slack or Microsoft Teams. They're sent a notification saying, “The owners X, Y, Z tagged you on this comment.”
You click on that link in your email or your text, and you're straight away onto the video. You're seeing that comment, and you know exactly what they're talking about, and you can resolve it. You can take multiple angles and stitch them together for these purposes. You're like, “Here's the angle from the front. Here's the angle from the back. This is what's happening. There was a customer altercation. Let's look at multiple angles. What was the customer doing prior to this?” You can stitch the whole narrative into one place and, and collaborate very seamlessly.
I like that opportunity to collaboration part and making it easy. Is there any audio that's captured?
Audio is something that is sensitive. There are states which do not allow audio capture. There are states that do. We provide audio capture, but the default habit is off. As an owner, we allow you to turn it on and we give a warning saying, “These are the states in which you should not be turning it on. It's not legal. If your camera's in other states, it's up to you then turn it on because,” then you're capturing conversations, which might be private, like two customers talking in a restaurant or something.
I was thinking of that altercation part. You want to capture that, but you don't want to capture the lovebirds having their nice dinner in society. You're not interested in that, but it's a delicate balance as there are Alexa devices near me and everything else.
There are privacy concerns. You have to tread the water carefully on where you want to take the technology. We are very concerned about which aspects would lead to privacy invasions and then having clear warnings built into it, or certain features not at all built into the product because those are places that can get businesses into trouble or ethically cross the line.
I do want to go back to theft for a second. Has it been useful in any use cases where an incident has been identified?
We heard a couple of different stories from our customers, and obviously, not every story makes it back to us. One of the restaurant owners deployed the technology before even the restaurant's opening while the restaurant is on the build phase. They realized that there was some material theft between what the contractor was saying and what was happening on the ground, and they could spot that.
Every time they open a new location before construction begins, they want the cameras because there are a lot of temporary workers coming in and out, and they're slowly bringing in more and more supplies and goods. The second example was they used to have these expensive wines on the wine shelf. One of their staff members is taking some of that off without accounting for it. They were able to find that and take care of it.
Addressing Theft In The Restaurant Industry
Especially in the restaurant industry, there are no secrets that I'm revealing right now. It's an issue that happens in the restaurant business. People will have a tendency over time to take advantage of what doesn't get managed and doesn't get watched and paid attention to. Knowing that there's a watchful eye will in and of itself, help detract from some of that systematically. The other thing that came to my mind is, I'm sure since it's AI and tech, and a lot of times with tech, we use 5% of what it can do. What are some of the other filters? What are some of the things that you can do?
There's more and more we are launching. One of the other things that our restaurant owners have found useful is very often, they have a a host who will greet you at the entrance of the restaurant. If you're a quick service restaurant, you have the people behind the counter who are making your sandwiches, burgers or whatever the case might be. Sometimes, the person is not present behind the counter because they have taken a phone call or they have gone away and they're not present. Lineups happen. We allow for people's presence counting. You can draw a box behind this, let's say host stand and say, “Between 5:00 PM to 9:00 PM, which is my busy hours, I want this person to be standing here, let's say 90% or 95% of the time.”
You can see variations in a few multiple stores saying, “The person is behind in the dinner time around 80% of the time, and this is 90% of the time.” Does that lead to lesser wait times and high revenues? The same thing behind the counter. You can draw zones and get whether your staff is present in those zones to serve the customers or not and you can get that data across multiple stores and you can compare them with bars, graphs and charts within a product then you can see, does that correlate to your revenue numbers or to customer happiness numbers or set reviews of different restaurant that does these things if your person being present and helpful. That's one.
This is something that is coming out. I'm excited about it. It may not be relevant to all restaurants, but I think it might be for a lot of restaurants is with license plate recognition, what you can do is start to understand who your repeat users. With that, you can start building and understanding who your loyal customers are.
Identify and reward your most loyal customers with license plate recognition technology. Offer personalized experiences to keep them coming back.
Eventually, you can set up a system where you get an alert or a notification when it this person has come five times to your restaurant in the last drive-through, a quick service restaurant or a normal regular restaurant and then you can be like, “If this person has come five times, let me gain their loyalty. Let me offer them something. Let me treat them a little better because they already like our food. There's a way I can keep and retain their business.” There are things like those that we're thinking through like, “How do we give you that data?”
Customer Interaction And AI Accuracy
Question on the license plate. Initially, I'm thinking drive-through, but you could also use that for a general parking lot. We've talked about it in other shows and on the website over the years, knowing who your customers are. People oftentimes believe they know more about their customers than they do. The data is always revelatory. I don't know if there's more filter stuff. I did have a follow-up. Have you used the tech yet to watch me as a customer? I'm trying to check out, or I'm looking for another glass of water and I'm doing the, “Where is my person?” thing. Is there a way to identify that and flag it somehow to the manager?
The moment you get into this technology, look at the person moving to the left and to the right. The problem is that each restaurant is built slightly differently and has different lighting and stuff. This is where AI starts to get a little fuzzy with false positives and false negatives. You can find AI demos online, which is like, “We can do it.” If you talk to most people who've experienced the product, they'll be like, “It's a hit and miss.” Once it's a hit-and-miss above a certain threshold, people start believing the data. We like to go for things that we have 95% plus certainty of getting right.
If the confidence intervals are lower than that, then we know that it's a matter of technology catching up. It's a matter of when and not if, but you have to intersect it at the right time to make it happen. We went beyond filters and back to what we started, what the owners find useful. The one thing that keeps surprising with all this tech and where AI can go is the way the traditional video recorders were built, they were built as silos.
If you talk to somebody who owns ten restaurant, they will tell you that, “To access my 10 video recorders, I probably have 10 different IP addresses or 10 different websites sometimes to go to.” Even if they are on the same platform and have planned it in such a way that they're all on the same camera system, they have to go between locations.
I have to go from an Arkansas location to a location in Ohio or something. With the fact that spot allows all the cameras in one platform, and visual is still one of the most powerful, meaning, if you close your eyes, it said that about 85% of all the data that we consume as human beings comes from our eyes. If you close your eyes, you're consuming that much less. When we show people videos of their sites, they can pull up their phones and see the floors across their six restaurants, and that gives them so much insight. They're like, “I didn't even know this was happening. Look at this table in the back. That's not how I want my customers to feel.”
Visual data is king in restaurant management. Spot's video intelligence platform empowers owners with candid insights across their locations, driving meaningful improvements.
There are these little things that, as owners, you care about, about the experiences that you see. You have these a-ha moments, which if you ask any restaurant owner that, “When you are on-site, do you get to see more about what's happening in the operations? Do you have five suggestions?” Things are a little more prepped up to when they go on-site because the owner is coming. This way you have a lot more candid view of what's happening across the restaurant. They continue to find that powerful, like most restaurant owners who use us, will have a video wall built on their phone or a monitor in the office, capturing these floor videos from their restaurant chains.
I do find a lot of power in those little details that you could bring to the next meeting, “We noticed this, and if we did that, it could lead to this. If you keep adding those over the course of time, it can be impactful.” As we start to close, I want to ask you about any limitations on geography if I were a restaurant on the East Coast, West Coast, North or South?
We have restaurants across the country using us. There’s no limitation with geography, we have no limitations on camera systems. You don't have to buy or install our cameras. We give cameras away for free. If you have any existing camera brand, we will work with them. That's another thing that restaurants like because when they build separate restaurants, they choose a local security vendor and they went with whatever cameras they recommend.
Very often, that means across the chain, they have three or four different systems and the cost of buying the same system means replacing cameras across that's expensive, time-consuming and operationally heavy. The fact that we don't need to change anything, we work with all camera brands, you deploy a box on the site and it then works on your phone and your desktop and everything else is super helpful.
Readers, go to Spot AI. Check them out. That was Rish Gupta, CoFounder and head of product there at Spot AI. Good stuff. I always love learning about this tech. It's amazing where it's going. We have to check in with you in 1 year from now or 6 months. It's probably fast. I know you guys are growing, you got funding and are doing all sorts of fun stuff. Good stuff. I appreciate you there. You can go to Spot AI to find out more. For more great restaurant marketing and service and people and tech tips, stay tuned to us here at RunningRestaurants.com. We'll see you next time. Thank you so much.
Thank you.
Important Links