Anomaly Detection in Video & Image Classification

We’re seeing and doing all sorts of interesting work in the image domain. Recent blog posts, white papers, and roundtables capture some of this work, such as image segmentation and classification to video highlights. But an Image area of broad interest that, to this point, we’ve but scratched the surface of is Video-based Anomaly Detection. It’s a challenging data science problem, in part due to the velocity of data streams and missing data, but has wide-ranging solution applicability.

In-store monitoring of customer movements and behavior.

Motion sensing, the antecedent to Video-based Anomaly Detection, isn’t new and there is a multitude of commercial solutions in that area. Anomaly Detection is something different and it opens the door to new, more advanced applications and more robust deployments. Part of the distinction between the two stems from “sensing” what’s usual behavior and what’s different.

Anomaly Detection

Walkers in the park look “normal”. The bicyclist is the anomaly. 


Anomaly detection requires the ability to understand a motion “baseline” and to trigger notifications based on deviations from that baseline. Having this ability offers the opportunity to deploy AI-monitored cameras in many more real-world situations across a wide range of security use cases, smart city monitoring, and more, wherein movements and behaviors can be tracked and measured with higher accuracy and at a much larger scale than ever before.

With 500 million video cameras in the world tracking these movements, a new approach is required to deal with this mountain of data. For this reason, Deep Learning and advances in edge computing are enabling a paradigm shift from video recording and human watchers toward AI monitoring. Many systems will have humans “in the loop,” with people being alerted to anomalies. But others won’t. For example, in the near future, smart cities will automatically respond to heavy traffic conditions with adjustments to the timing of stoplights, and they’ll do so routinely without human intervention.

Human in the Loop

Human in the loop.

As on many AI fronts, this is an exciting time and the opportunities are numerous. Stay tuned for more from, and let’s talk about your ideas on Video-based Anomaly Detection or AI more broadly.

Artificial Intelligence Will Take Jobs

A few months back, Treasury Secretary Steve Mnuchin said that AI wasn’t on his radar as a concern for taking over the American labor force and went on to say that such a concern might be warranted in “50 to 100 more years.” If you’re reading this, odds are you also think this is a naive, ill-informed view.

An array of experts, including Mnuchin’s former employer, Goldman Sachs, disagree with this viewpoint. As PwC states, 38% of US jobs will be gone by 2030. On the surface, that’s terrifying, and not terribly far into the future. It’s also a reasonable, thoughtful view, and a future reality for which we should prepare.

Naysayers maintain that the same was said of the industrial and technological revolutions and pessimistic views of the future labor market were proved wrong. This is true. Those predicting doom in those times were dead wrong. In both cases, technological advances drove massive economic growth and created huge numbers of new jobs.

Is this time different?

It is. Markedly so.

The industrial revolution delegated our labor to machines. Technology has tackled the mundane and repetitive, connected our world, and, more, has substantially enhanced individual productivity. These innovations replaced our muscle and boosted the output of our minds. They didn’t perform human-level functions. The coming wave of AI will.

Truckers, taxi and delivery drivers, they are the obvious, low-hanging fruit, ripe for AI replacement. But the job losses will be much wider, cutting deeply into retail and customer service, impacting professional services like accounting, legal, and much more. AI won’t just take jobs. Its impacts on all industries will create new opportunities for software engineers and data scientists. The rate of job creation, however, will lag far behind that of job erosion.

But it’s not all bad! AI is a massive economic catalyst. The economy will grow and goods will be affordable. We’re going to have to adjust to a fundamental disconnect between labor and economic output. This won’t be easy. The equitable distribution of the fruits of this paradigm shift will dominate the social and political conversation of the next 5-15 years. And if I’m right more than wrong in this post, basic income will happen (if only after much kicking and screaming by many). We’ll be able to afford it. Not just that — most will enjoy a better standard of living than today while also working less.

I might be wrong. The experts might be wrong. You might think I’m crazy (let’s discuss in the comments). But independent of specific outcomes, I hope we can agree that we’re on the precipice of another technological revolution and these are exciting times!

Programmatic Video Highlights

AI Video Classification

For many years, and with rapidly accelerating levels of targeting sophistication, marketers have been tailoring their messaging to our tastes. Leveraging our data and capitalizing upon our shopping behaviors, they have successfully delivered finely-tuned, personalized messaging.

Consumers are curating their media ever more by the day. We’re buying smaller cable bundles, cutting cords, and buying OTT services a la carte. At the same time, we’re watching more and more short-form video. Video media is tilting toward snack-size bites and, of course, on demand.

Cable has been in decline for years and the effects are now hitting ESPN, once the mainstay of a cable package. Even live sports programming, long considered must see and even bulletproof by media executives, has seen declining viewership.


So what’s to be done?

To thrive, and perhaps merely to survive, content owners must adapt. Leagues and networks have come a long way toward embracing a “TV Everywhere” distribution model despite the obnoxious gates at every turn. But that’s not enough and the sports leagues know it.

While there are many reasons for declining viewership and low engagement among younger audiences, length of games and broadcasts are a significant factor. The leagues recognize that games are too long. The NBA has made some changes that will speed up the action and the NFL is also considering shortening games to avoid losing viewership. MLB has long been tinkering in the same vein. These changes are small, incremental, and of little consequence to the declining number of viewers.

Most sporting events are characterized by long stretches of calm, less interesting play that is occasionally accented by higher intensity action. Consider for a moment how much actual action there is in a typical football or baseball game. Intuitively, most sports fans know that the bulk of the three-hour event is consumed by the time between plays and pitches. Still, it’s shocking to see the numbers from the Wall Street Journal, which point out that there are only 11 minutes of action in a typical football game and a mere 18 minutes in a typical baseball game.


A transformational opportunity

There is so much more they can do. Recent advances in neural network technology have enabled an array of features to be extracted from streaming video. The applications are broad and the impacts significant. In this sports media context, the opportunity is nothing short of transformational.

Computers can now be trained to programmatically classify the action in the underlying video. With intelligence around what happens where in the game video, the productization opportunities are endless. Fans could catch all of the action, or whatever plays and players are most important to them, in just a few minutes. With a large indexed database of sports media content, the leagues could present near unlimited content personalization to fans.

Want to see David Ortiz’s last ten home runs? Done.

Want to see Tom Brady’s last ten TD passes? You’re welcome.

Robust features like these will drive engagement and revenue. With this level of control, fans are more likely to subscribe to premium offerings, offering predictable recurring revenue that will outpace advertising in the long run.

Computer-driven, personalized content is going to happen. It’s going to be amazing, and we are one step closer to getting there.

Voice Ordering != Voice Shopping

Voice Ordering is Here

Voice shopping is coming, and it’s far more interesting.

Siri has been with us for years, but it’s in the last few months and largely due to Amazon that voice assistants have won rapid adoption and heightened awareness.

Over the last few months, we’ve been shown the power of a new interaction paradigm. I have an Echo Dot and I love it. Controlling media and some lights are the most useful applications so far. The Rock, Paper, Scissors skill… yeah, that one’s probably not going to see as much use.

But let’s not forget that this slick device is brought to us by the most dominant ecommerce business in the known universe. So it’s great for voice shopping, right? No. Not at all. It doesn’t do “shopping” at all.

But I heard the story about the six-year-old who ordered herself a dollhouse? So did I, and it reinforces my point. Let me explain.

The current state of commerce via Alexa is almost like a broad set of voice-operated Dash Buttons. For quick reorders of things you buy regularly and when you’re not interested in price comparisons, it’s fine. What it’s not — voice shopping.

Shopping is an exercise in exploration, research, and comparison. That experience requires a friendly and intelligent guide. As such, voice shopping isn’t supported by the ubiquitous directive-driven (do X, response, end) voice assistants.

Shopping is about feature and price comparison, consideration of reviews, suggestions from smart recommendation engines, and more. Voice shopping is enabled by a conversational voice experience, one that understands history and context, and delivers a far richer experience than is widely available today.

The Mobile Impact

Mobile commerce isn’t new and is still growing fast. But despite consumers spending far more time on mobile devices than on desktops (broadly defined, including laptops), small screen ecommerce spending still lags far behind.

So why can’t merchants close on mobile? The small screen presents numerous challenges.

Small screens make promotion difficult and negatively impact upselling and cross-selling. Another major factor, one you’ve probably experienced, is the often terrible mobile checkout process. Odds are, you’ve abandoned a mobile purchase path after fiddling with some poorly designed forms. I have. Maybe you went back via your laptop. Maybe you didn’t. Either way, that’s terrible user experience.

Through voice, retailers can now bring a human commerce experience to the small screen. It’s a new, unparalleled engagement opportunity; a chance to converse with your customer, capture real intelligence about their needs, and offer just the right thing. It’s an intelligent personal shopper in the hands of every customer.

Come reimagine voice shopping with us. Imagine product discovery and comparison, driven by voice. Imagine being offered just what you were looking for, based on a natural language description of what you need. Imagine adjusting your cart with your voice. Imagine entering your payment and shipping info quickly and seamlessly, via voice. It’s all possible and it’s coming soon.


Watson’s Reckoning

Watson’s Reckoning

To most in the know, IBM’s Watson has long been considered more hype and marketing than technical reality. Presented as infinitely capable, bleeding edge technology, you might think the well-known Watson brand would be delivering explosive growth to IBM.

Reality is far different. IBM’s stock is down in a roaring market. The company is, in effect, laying off thousands of workers by ending its work-from-home policy. More than $60M has perhaps been wasted by MD Anderson on a failed Watson project. All of this is happening against the backdrop of a rapidly expanding market for Machine Learning solutions.

But why? I saw Watson dominate on Jeopardy.

And dominate it did, soundly beating Ken Jennings and Brad Reuter. So think for a moment about what Watson was built to do. Watson, as was proven then, is a strong Q&A engine. It does a fine job in this realm and was truly state of the art…in 2011. In this rapidly-expanding corner of the tech universe, that’s an eternity ago. The world has changed exponentially, and Watson hasn’t kept pace.

So what’s wrong with Watson?

  • It’s not the all-encompassing answer to all businesses. It offers some core competencies in Natural Language and other domains, but Watson, like any Machine Learning tech, and perhaps more than most, requires a high degree of customization to do anything useful. As such, it’s a brand around which Big Blue sells services. Expensive services.
  • The tech is now old. The bleeding edge of Machine Learning is Deep Learning, leveraging architectures Watson isn’t built to support.
  • The best talent is going elsewhere. With the next generation of tech leaders competing for talent, IBM is now outgunned.
  • …and much more discussed here.

The Machine Learning market is strong and growing. IBM has been lapped by Google, Facebook, and other big-name companies, and these leaders are open sourcing much of their work.

Will Watson survive? Time will tell.

Mark Zuckerberg, Global Editor-in-Chief


Mark Zuckerberg, Global Editor-in-Chief

Not a hot take:  Facebook is a media company


Not just a “social” media company. Simply put, they’re the world’s biggest media company.

What about Google?  Not a media company.

But Google Plus?  Irrelevant.  Google News?  Curated differently, and far less — professional media only, promoted based on preferences and relevance.


So why Facebook?  Curation.

The day Facebook started curating our feeds is the day it became a media company. I’m not complaining or even suggesting there’s an alternative. Whatever the underlying exact metrics that govern our feeds, they are critical to our use of the platform and FB would be a noisy shit-show without these smart, useful measures.

The fact that these decisions are being made by computers and on the fly doesn’t absolve Facebook of editorial responsibility. The algorithms report to the engineers and the engineers to Mark. So now, as has been the case for a while but was forcefully exposed this fall, he’s got significant editorial responsibility.

Facebook stands alone in its reach, relevance, and responsibility. Mark Zuckerberg is now the world’s Editor-in-Chief.

Think that’s hyperbole? Not with 1.8B MAUs. Not with a market cap over $300B. And not when you’re the founding CEO who is the face of and wield significant shareholder voting control over, the company.

This isn’t simply about fake news, or silos, confirmation bias bubbles and the like. It’s much bigger than what’s trending, how, and why. As the world’s preeminent news organization, Facebook is going to have to figure out all of this and more.

This is a huge, complicated problem. Balancing their business objectives and this enormous responsibility will be difficult, but it’s in their interest, and their customers’ interests, to make the necessary investments in this area. Fortunately, they have billions in cash and many smart people on the team.


So Mark — years ago, you probably did imagine yourself in Bill Gates’ shoes. You’ve done that. Awesome. Congrats. Now welcome to a whole new level of responsibility you may never have considered.

Best of luck. The fourth estate may depend on it.

A Tale of Two Companies

A Tale of Two Companies

Having wound down my startup, I’m on the hunt for a new job. As many of you know well, the whole process sucks.

I’d like to relate (and offer commentary on, of course) a couple of recent experiences with Boston area startups. It’s a quick peek into the true culture of these companies and offers immediate insight into where you might want to work, or NOT.


Company 1 — Drift

Role:  Sales          Fit for my experience:   Very Good

A customer-centric business founded by some ridiculously successful entrepreneurs. Given their success and likely ability to draw talent from wherever they choose, they could take an aloof approach to recruiting and hiring. They don’t.

Contacting Drift was unexceptional, except to the extent it was human. I emailed the recruiter. He got back to me within a couple of hours to report that they’d filled that role, but he’d like to stay in touch on future opportunities.

Fast. Professional. Honest. Warm.

I doubt it took him more than a couple of minutes to create just a little more good will toward the company. I expect he does this every day. I expect everyone there does this every day. Why? Because it’s the right thing to do as people and as a business. As a startup business, it’s not an overwhelming amount of work and simply the right thing to do.

I don’t know whether I’ll ever be under real consideration for a role at Drift, but I’m certain that, culturally, it’s a place I’d want to work.


Company 2 — let’s call them BevSpot

Role:  VP, Sales          Fit for my experience:   Fair

SaaS. Rookie founders.

Per their process, I emailed careers@. As a sales pro, I use Streak in my Gmail to track opens. I know my thoughtful, professional, and job and company-specific cover letter email was opened. I assume my resume was viewed. But there was no response from BevSpot, not even the usual “Thanks. We got it. We’ll be in touch if appropriate.” That’s a reasonable minimum expectation.

Applicants deserve to know that they’re in the queue, not simply in a black hole. Further, a startup making its first executive hire owes thoughtful, professional applicants a professional response. Tell me to pound sand. I can take it. That’s far better than nothing.

After hearing nothing for three weeks, I followed up. It wasn’t a “please interview me” email.

I know someone read my email — Liz? Courtney? — because I’m a sales pro who tracks even some personal email.
BevSpot is growing up and this VP Sales hire needs to be an experienced, professional adult. It’s time for your hiring practices to reflect the company’s maturation.
At a minimum, you owe applicants, at least the serious ones, a basic acknowledgement that their email / resume was received. That’s the minimum. Fortunately, you can implement that today.
I don’t expect this will help my candidacy. (Regardless, I’ve done this before — helped SaaS businesses grow, and grow up. I’m ready to do it again, but convincing you to interview me isn’t my aim here.) This is about helping you build and defend your brand. Maybe you can just ignore fresh grads. I wouldn’t, but there’s less harm in it, I suppose. With an expectation of success, the VP Sales role won’t be your last experienced hire. Professional adults just want to be treated like… professional adults.

No, I didn’t expect this would help me get the job, but it felt good to vent a little, I believe I’m 100% in the right, and maybe I can help them treat future applicants better.

This email was opened within hours. Still nothing from BevSpot. About 100 employees, two people working in People Operations, and recruitment is already broken.

Weeks later, and after noticing that the role had been filled, I emailed the CEO to relay my experience, which I viewed as unprofessional. He responded apologetically within 12 hours. Still nothing from People Operations. Hacks.

This is not a place you want to work. This isn’t sour grapes, I promise. This is an organization that, despite being early stage and small, already has a culture problem. They treat people like shit, and you don’t even have to take my word for it.


One of these companies clearly is and will continue to be, a great place to work.

One of these companies will be successful.

The other? Without a cultural enema, not so much.

Yes, the job hunt sucks. But sometimes there’s valuable insight buried inside the frustration. And this much insight was available without even interviewing.


*** Update 12/16/16:  Broken culture. Broken company.

The Story of No Pots & Pans

No Pots & PansThe Story of No Pots & Pans

No Pots & Pans produced, sold, and delivered healthy, prepared meals in the Boston area and MetroWest.

I developed the concept, then built and launched the business, bootstrapped and by myself.  I did it all — marketing, web dev, cooking, legal, and finance. I built the AWS ecommerce site and marketing stack. I set up a Delaware C-Corp and filed the taxes. I modeled, tracked, and managed everything; finance to marketing. I built a brand with a highly engaged audience across the website, email, and social.

We offered 3-4 different meal selections weekly, trying to appeal to a wide range of tastes. We cooked, chilled, packed (in compostable, fiber containers), and delivered meals to customers, most often to coolers they would leave out for us.

The business grew from $0 to almost $10,000 in monthly revenue in just a few months. We sold thousands of meals to hundreds of happy, loyal customers. Working with local food banks, we provided hundreds of meals to those in need.

No Pots & Pans


Proud Partner and Supporter of Local Food Pantries

No Pots & Pans provided hundreds of meals to people in need, through food donations and per order cash contributions.

Every order placed, for every day we were open, provided a meal to someone in need. In partnership with the GBFB, we donated $.33 per order.

Most weeks, we overproduced meals. When we did, I’d bring the extras and any excess inventory to the local food pantry. I heard that the meals went quickly and were well received.


But it’s not all rosy, or I’d still be at it. It’s a difficult business, logistically challenging and it doesn’t enjoy the high margins of a technology business. Unable to make meaningful investments in customer acquisition, I didn’t have a near term path to financial rewards commensurate with the 80-100 hour weeks I was putting in. In October of 2016, I made the difficult decision to pull the plug.

As most entrepreneurs will report, this was the hardest, most professionally fulfilling experience of my career. My story is unremarkable and ends with the business closing within a year of launching. But there’s a story, nonetheless, and it’s not one of failure. It’s an honest take from inside the grind. If you’re interested, scroll with me.


The Opportunity

Market inefficiency:

People want to eat healthier.

Home cooking is healthier, but…

…it’s near all-time lows and declining.

The ubiquitous delivery options (pizza and Chinese) are wildly unhealthy.

The No Pots & Pans business:

Make healthy, tasty meals.

Sell them online.

Delight customers.



Takeout and delivery combined are almost $100B annually and growing. The Boston market alone is about $2B. The concept is simple and broadly similar to what Munchery seemed to be doing successfully in California. (Now, perhaps, we know a little differently.)


Making it Happen

Nothing happens by accident. This was a combination of my personal passion for cooking with professional experience in marketing and e-commerce. To the delight of my family, as it solved a real-world problem in our house, I launched No Pots & Pans late in 2015.

Starting any business is hard. Starting a food business; working through the permitting process is harder.

I had hoped to work with a partner who would produce the meals, a caterer or restaurant, and sell them to me at a fair price. While exploring this route, I learned that if I were to buy meals from another company and deliver them, I’d need to do so with a commercial refrigeration vehicle. With last mile delivery and logistics, generally, already expected to be an expensive challenge, this requirement broke that model. Back to the drawing board.

The local food delivery businesses you know don’t face these requirements. Why? Because they operate their own commercial kitchen. “Chain of custody”, for lack of a better explanation, matters. When you produce the food, you can deliver it in normal cars, as an extension of your food establishment permit. All I need now is a commercial kitchen.

Bootstrapping the business, I couldn’t afford a commercial kitchen or even a shared kitchen rental in the first couple of months. I also now need kitchen staff. I can cook, but well enough? I have some restaurant experience from 15+ years ago, but I’d rather focus on growing the business. Off to find a kitchen and staff… maybe I can offer services to offset my costs.

After reaching out to most caterers in the Boston area and being quickly rejected by several, I got a bite. I found a caterer whose business was struggling and where I could add some immediate value in her operation. At least conceptually, she could help me meet both my compliance and operational needs around a kitchen and a staff. This is what I needed, so I immersed myself in helping her floundering operation stay afloat.

First, we need to turn the kitchen into a shared kitchen. This is the first one of its kind in the city, so I need to draft a set of SOPs and associated rules. For all who follow, you’re welcome. With the plan blessed, we dove into inspection processes, though slow and expensive for a fast moving startup operating on a wing and a prayer, it went smoothly enough. On paper, I built a fully licensed, functioning food establishment (I like the term “virtual restaurant”), in about 60 days and for $500 and some consulting sweat equity.


Open for Business

No Pots & Pans open-for-business

We launched in November 2015, offering delivery just a couple of days a week.  This allowed us to cook for a few hours, chill and pack the food, then deliver it over the next couple of days while it was still fresh.

I’m not a chef, so I sought professional help with the menu planning and food production.  At launch and for the first few weeks, I expected this to be a losing proposition on low output volume and sales.  Keen to control costs and get my hands dirty in the business, I dove into the kitchen work, acting as prep cook and dishwasher for our first production run.  We’d done some test cooking in the weeks prior; dry runs for practice, menu planning, and website photos.  That was fun, but the real thing, the production runs where the food was going out to my customers (ok, so most of the initial customers were friends and family) was inspiring.


Unfortunately, the folks I paid to lead and execute the food production proved expensive and highly inefficient. More than just the hours and dollars spent cooking, it was painfully apparent that there was no plan, little process, and things should have moved faster. Further, while this disappointed still existed only in my head, the leaders of the team that I hired requested more money — payment for extra hours. Though the ask was small in dollar terms, it was the final straw. I paid them what was requested, thanked them for their help, and anointed myself, Head Chef. From Week 2, forward, I did the menu planning and led meal production.

It was hard. The cooking was typically a mad scramble, but that was only a few hours a week. It was everything else surrounding the food operation that constituted a full-time job, in addition to running the business and fighting like hell to grow it.

Saturdays became my test cooking days. I’d make a few meals, track the inputs and quantities, sample the food, and take pictures. The winners went into my “meal library” or directly to the site for sale. Sunday, the new menu went up. Monday, purchase planning, informed by orders and demand speculation, began. Tuesday was my food shopping day, often taking me to Restaurant Depot, a supermarket, looping from home to Somerville, filling the car and spending hundreds of dollars buying produce and meats by the case. Wednesday morning was a mad dash to the kitchen to crank out meals for delivery that very afternoon. Thursdays and Fridays were delivery days. Cruising around, doing deliveries on those days was the most relaxing part of the week.


Early returns were positive. Customers said they loved the convenience and the food. The reorder rate was above 70%, with many customers incorporating us into their weekly routine and ordering every week.

After just a few weeks in business, my press outreach started paying dividends. After speaking with a reporter for the Hudson Sun (local paper that rolls up to the MetroWest Daily News), I was expecting a small article in the Sun. About ten days later, I wake up to a series of customer emails and see my Google Analytics blowing up (when you’re accustomed to 0-2 people on site at a time, 10-20 users counts as blowing up). We had earned a half page feature in the MWDN print edition, along with home page placement on the website. The newsletters they sent boosted traffic even higher.

Media being an echo chamber, we were soon featured in BostInno and Boston Magazine. The earned media was valuable and came at just the right time. In one week, strongly supported by this exposure, sales doubled.

Sales doubling in the span of 48 hours is a great problem to have. When you’re in a business that makes real world, physical products, it’s an enormous challenge. When your product is perishable and your customers want it within 12-48 hours, there’s a great deal of pressure.

As a greedy capitalist and hungry entrepreneur, I fought like hell to take all orders and avoid selling out. Watching the orders roll in, new customer upon new customer, was both a thrill and incredibly stressful.

We were stretched, depending on your perspective, to or beyond our capacity. I’d be lying if I failed to acknowledge that quality suffered some, but we took and fulfilled every order. It was, by far, our best week to that point – 2X any prior week.

I solicited help from friends and family, who came through in significant roles, and I couldn’t have pulled it off without their impactful contributions. Thank you.

News article that includes “Hudson man” and my picture. Could be worse, right?


The Grind

I hired a couple more prep cooks. I started interviewing delivery drivers. We were off to the races. Prior to our big week, everything pointed up and to the right. We should continue growing, and on a compounded basis against a larger customer base, right? Not so fast.

Despite making some small investments in advertising and continuing to throw myself into content production, deploying a referral program, and making personal, individualized appeals with discount offers, we did not recapture the momentum of that week in January. Taking a longer view, there was an argument for growth — the trendline was there, with an upside outlier. Still, it was a slow grind higher, week by week, if there was any growth at all.

Then summer hit. I had expected it to be slow. I had not expected July to be off 50% and August more than that. In hindsight, now understanding the seasonality, this is a business that could shut down in the summer. Lesson learned.

With an eye toward the fall, I tried to remain optimistic. I had high hopes for September and, if the business was to continue, I needed to see a trajectory that would allow me to personally make money from the business.

In parallel, and going all the way back to pre-launch, I sought investment. In the current “on-demand” funding climate, I understood that it was highly unlikely I could attract institutional seed capital until I had strong, steady, and growing revenue. Unit economics were solved Day One. Customer economics looked outstanding, but it’s hard to know how much churn would have ticked up in a significantly expanded customer base. Regardless, I was unable to bring in investors and the capital I felt I needed to fuel growth.

September, by any measure, was strong. It was one of our best months in business. But it wasn’t enough. I needed to see double what we did, in order to keep the lights on. As we turned the corner into October, I made the difficult decision to cease operations.

It was fun. I learned a lot. I’ll be back to do it again soon, but not in food and not B2C.