‹ View all briefings
July 31, 2020

Five Startup Ideas #009

Talking to animals with Machine Learning, Headspace for hypnosis, and tapping debt.

Photo by


While the country spent much of this week contemplating the scale of tech's giants, we're glad to turn our attention to those yet to bud. Ten new ideas today for those ready to become the next Bezos. (Let's hope none of us become the next Zuck.)

Last week's ideas got a strong reception with Ali Afridi, Molly O'Shea, and several others receiving responses to their submissions. Always exciting to see new connections made.

The group chat continues to be a source of insight and collaboration. This week, we talked about esports arenas, DTC CAC, infrastructure for politicians, the creator economy, and plenty more. Thank you to all who make it a lively, inclusive place.

We're always looking for more generous, thoughtful people to add. If you'd like to be a part of the community, hit the button below and tell us about yourself. If you think we've missed your application or you want to add to it, feel free to email. We definitely do our best to look at everything carefully, but this is very much a two-man operation :)

On to the quote of the week. Tim O'Reilly has been a crucial chronicler and interpreter of tech's rise. Here's what he said about invention:

An invention has to make sense in the world it finishes in, not in the world it started.

A good reminder of the foresight, risk, and vision it takes to make something new. Respect to all who take up the challenge. Now, onto to the ideas themselves. Lightly edited by us for the sake of clarity.

Deep learning for animal understanding

Applying deep learning to identify patterns in animal behavior

Deep learning is great at finding signal in noise. There's been more and more compelling research surrounding understanding things like pain or other communication/emotional patterns within animals using ML techniques applied to audio and video. You can envision a slope where we end up able to better understand animal health and preference through patterns, similar to how we can identify these things visually and aurally in humans (though this is controversial). I don't know where the exact business model is here (CPG? Healthcare? Dog friendship?), but I'm only partly joking when I talk about TAM creation/expansion if we can unlock the next level of animal understanding.

Michael Dempsey, Partner at Compound

Headspace for hypnosis

A modern, accessible hypnotherapy platform for clear minds and clear hearts (Ed: can't lose!)

Mental healthcare is the #1 public health crisis in America. Today, 38M Americans are on antidepressants, while 20M Americans are on anti-anxiety medication. With 46% of Americans experiencing mental illness during their lifetime, US annual spend on mental health treatments and ancillary services has reached $238B. Today’s treatments are far from sufficient.

Enter hypnotherapy. Hypnotherapy, which induces a state of intense concentration or focused attention, could clear the way for deeper psychological processing and acceptance. Research shows strong evidence surrounding the potential for hypnosis to treat pain, irritable bowel syndrome, PTSD, and insomnia while an emerging body of study suggests hypnosis might also be applicable for depression, anxiety, smoking cessation, and weight loss.

Might we be able to bring such world positive change to the masses through a digital application or experience? What meditation is today, hypnotherapy might be tomorrow.

Meera Clark, Investor at Obvious Ventures

The self-actualization bundle

A personal advisory board for every pillar of well-being

There is a massive impact gap in the consumer wellness market. Spend is increasing rapidly, but results are hard to come by. We struggle to stay fit, manage our finances, and maintain healthy relationships (among a plethora of other disappointing shortcomings). Content and one-size-fits-all advice abound, but truly personalized experiences that drive adherence and results are largely inaccessible. The complexity and costs associated with assembling a support team leave it off-limits to all but the very highest earners.

A company could bundle access to an "advisory board" of personal coaches, counselors, and educators, pairing expertise with tools that allow practitioners to scale. Such a solution could significantly bend the cost curve of access down while helping tens of thousands of practitioners build thriving small businesses.

Brett Bivens, Investor at TechNexus and Creator of Venture Desktop

Your health OS

Unified "smart" health metrics aggregator

Apple Watch to track fitness, Oura Ring to track sleep, Whoop Band to monitor strain, Qardio for blood pressure — we each have a heterogeneous array of trackers monitoring various aspects of our health, each with a different dashboard. There's no way to monitor all the metrics that matter, making it harder to get a picture of your overall health.

What if there was a health aggregator platform that ingested data from all tracking devices, providing a snapshot of overall health? This device could then be paired with software that ran predictive analytics, creating a model of what your levels were today and what they should be tomorrow. Anomalies could be flagged and potentially shared with a doctor. Beyond the metrics widely available already, it's easy to envision how such a system might encourage further quantification — think electrolyte consumption, stool inflammation, blood oxygen levels, or even the presence of COVID antibodies. Individuals with chronic diseases, like celiac, could begin to grasp how different behaviors impacted how they felt and performed. In the long-run, that could be used to enhance care.

There's reason to believe there's a market for such a system. Peloton and Mirror indicate growing consumer fitness spending, while at-home testing companies like EverlyWell, show the wisdom of bringing health and hardware directly to consumers. If the collected data were sufficiently robust, a Livongo-esque play might be possible — developing benchmarks and providing rich care management.

Nathan Baschez has a good take on why Apple is uniquely poised to build something like this (but hasn’t yet!), and Frank Chen describes Life in 2030, painting the picture of a world in which we track health metrics every day.

Grace Isford, Investor at Canvas

A wearable subvocalization device

"Talk" with a virtual assistant without needing to speak out loud

Voice assistants have mostly fallen short of their expected promise. Siri, Google, and Alexa are nice for small things, but most people don't really use them. One of the biggest barriers is the human behavior change required to feel comfortable talking out loud to your phone when other people are around. And of course, talking to your phone in a noisy environment isn't great either.

Instead, subvocalization devices worn on the jaw are able to leverage ML to understand the imperceptible mouth and tongue movements made when speaking to yourself, internally. This wearable allows you to "speak" without opening your mouth, allowing instructions to be passed silently to a device. It can be paired with headphones so you can receive responses.

You could use this in noisy environments, in quiet office settings — really anywhere.

The tech has been validated at a basic level in an MIT research project, and with further advances in language models, the ability to more accurately convert subvocalizations to text and understanding should be even more doable.

Touch-less and voice-less interaction could be the next step-change in computing interfaces if subvocalization devices are commercialized.

Hima Tammineedi, Founder of Stealth and former ML researcher at CMU

Founder-friendly debt terms

SaaS credit scoring with a marketplace

Buzz surrounding the future of startup financing has surged this year and is further accentuated by the tech industry’s antifragility in the public markets. Despite the relatively predictable nature of recurring revenue businesses, many banks and investors still don’t have the capability to lend to startups as their models rely on GAAP metrics instead of more relevant measures such as retention, LTV: CAC, sales efficiency, and so on. While there are now numerous lenders with dedicated venture debt teams, there is an opportunity to build an underlying startup-focused credit scoring model and sell it to less innovative lenders.

With the powerful new underwriting model in hand, the next step is a SaaS-enabled marketplace:

  1. On one side, you connect to a startup's internal systems to assess creditworthiness and assign a credit score. Startups benefit because the marketplace produces the most competitive term sheets rather than going to a few lenders for one-off requests.
  2. On the other side, banks and investors, armed with an underwriting model, can bid for a given startup. Lenders benefit by gaining exposure to a previously inaccessible asset class.

A flywheel would ensue where more transactions generate more data and more accurate pricing. The challenge is in gathering the initial data set to build an accurate, robust underwriting model.

Zach Bahm, Venture Debt at HSBC

Grubhub for in-person dining

App to order ahead at restaurants, particularly during COVID-19

Most restaurants in New York have transitioned from physical menus to QR codes. A service could be offered that further minimized interactions, allowing customers to order prior to arriving at a restaurant. With location services enabled, the restaurant could properly time service. Thanks to the tech developed by delivery apps, integrating with a restaurant's order queue is fairly straightforward.

Ultimately, I see such a solution as a "COVID play plus." The tailwinds of the pandemic could precipitate adoption — this service would accelerate turnover, minimize need for staff, and theoretically give customers peace of mind — but the model could be enduring. In the long run, customers will not want to lose the ability to pre-order food, given the obvious convenience and time-savings.

Thomas Mecattaf, Analyst at Charge Ventures

Headless identity and user monitoring

Continuing to power the developer economy

There has been a bit of scrum in recent weeks over how the next generation of SaaS will be developer-first. The basic logic is that data models, logic, and clients/UI for traditional business applications will be decoupled into "headless" solutions, where developers build on top of a robust API and configurable backend. Examples in e-commerce include CommerceLayer and in automation, Robocorp. This trend also falls in line with the JAMstack movement and ecosystem. I think it'd be interesting to expose user, identity, and account behavior data to developers through an API, similar to what Castle does. Then, they could build custom experiences for access policy, bot detection, and account takeovers to fight fraud and risk.

Aashay Sanghvi, Investor at Haystack

Stripe Atlas for cooperatives

Self-reinforcing, non-dilutive growth

With the rise of the “every startup is a bank” narrative, I’m intrigued by the “every startup is a credit union” opportunity.

Credit unions are cooperatives. There are 29K cooperatives in the US alone. They generate $500B+ in annual sales, hold $3.1T in assets, and employ 856K people who earn $25B in wages.

It’s a pretty straightforward model that’s hundreds of years old. Members — individuals or companies — pay a fee to become owners, benefit from quality, cheap goods or services, and receive a distribution of profits at the end of the year. The capital base grows over time and is used for expansion.

Straightforward, but antiquated to stand up.

This is the request: a Stripe Atlas-inspired method of organizing and deploying an outside-capital-ready organization centered on cooperative mechanics. That could be as a standalone new entity, or as an injection into an existing business.

For the underrepresented, the underserved, and the repressed, this is the flywheel for economic power. For the acquisitive, this is the best source of cheap capital that you own. You can imagine a Twitch cooperative committed to building and selling tools for streamers, or a teacher's cooperative that runs a distributed school.

Visa, Land O’Lakes, REI, Ocean Spray, Liberty Mutual, and dozens of other large companies operate as cooperatives — there's every reason to think this model can reach scale. I'm investigating building in this space and would love to talk to others that are interested. Please reach out!

Shlok Vaidya, OnDeck Fellow

Plausible dystopia

How 5G and "voice attribution" will kill conversation

This may be less of an idea and more of a warning. Or perhaps an invitation to burgeoning supervillains.

Have you ever discussed a product, then found it populating your Instagram feed moments later? That experience, and others like it, have contributed to 43% of Americans believing their phones are listening to them. Whether it is true or not, we have made our peace with some degree of surveillance. That's without mentioning the devices we ask to snoop on us like our Alexa or Google Home.

There's every reason to think that such surveillance will expand during the coronavirus pandemic and once 5G attains ubiquity. That may change cultural mores too — rather than resisting oversight we could learn to embrace it. Few inducements work as well as money.

I imagine an app that lives on your phone, and perhaps other listening devices, always-on, waiting for one thing in particular: recommendations. We still value the endorsements of friends more than information gleaned from other sources. Building on Alex Lieberman's RFS from last week — perhaps we should be paid for it.

In this world, I might tell my friend about an Airbnb I stayed at upstate, a jacket I bought, a book I read. The app would log all three recommendations and the human entity to whom they were suggested. If my friend went on to follow any such recommendations, even months later, some portion of affiliate revenue would pass to me, and perhaps be shared with others who reinforced recommendations over time. Essentially, an integrated system of "voice attribution."

On one end, it might provide passive income. On the other, we would observe the death of conversation. Each of us using our allotted moments to sell, sell, sell until language is fully mercantilized.

Mario Gabriele, Investor at Charge and Editor of The Generalist

Thank you for reading. If you have someone you'd like to hear an idea from, let us know. We'll do our best to get them into the mix. We'd love to hear your ideas, too.