Name: Khushang Hirpara

Business: Using AI to Create Personalized Supplements – CLD-9

CU Graduate Class of 2022, Major in Entrepreneurship and Management

Tell me a bit about yourself and what you’re currently working on.

I’m an award-winning founder and operator in Boulder that specializes in a variety of fields such as Design, Product Development, and Sourcing. But truly, I am obsessed with using leading technology and research to build products that make people’s lives easier. I’m currently working on addressing the complexity and risk involved in researching and using supplementation by allowing users to simply have a conversation, create their supplement formulation/routine, and drink it. Believe it or not, there’s an insane amount of AI research and robotics involved, which I’ll talk about later.

“What was your major and what led you to choose that major?”

I was an entrepreneurship and management grad at Leeds that graduated a year early. I either had the option of pursuing computer science in Seattle, Washington, or going to Boulder for their entrepreneurship program. Unfortunately, I couldn’t see myself in the comp sci environment at Seattle, and I had done a ton of research on CU’s Startup program. From the start, I had planned out my time at CU to create solutions, graduate early, and spend a year trying to scale a venture.

“Have you been able to make any connections at Leeds that have helped you with your business?”

I’ve made a ton of connections and gained some amazing insights from Leeds I haven’t had from any tech spaces in Washington. However, all of these connections were entirely from guest speakers from classes. I deeply encourage all students to connect with guest speakers whenever they speak at the school. It’s the easiest way to make lifelong connections with influential people that have done what you’re trying to do.

“Obviously, balancing school and your own venture are not the easiest things to do. What’s your strategy to manage both?”

It was nearly impossible to try to balance both. A startup is a time-consuming thing. The likelihood of your success is already low even if you’re spending 8 hours a day on your startup. Imagine mixing that with school. I had to get really resourceful to keep my grades above a 3.6. I’d essentially live at the Leeds conference rooms. They don’t know this, but there were days where I would sleep and work there for 48-72 hours at a time. Nonetheless, I ended up spending the minimal amount of time needed for school, which was around 3-4 hours a week. My teachers were also really forgiving; they knew what I was working on after winning NVC (new venture creation) and gave me a lot of free room to submit things late.

“Do you think, knowing what you know now, you’d have taken the same path?”

Yes and no. I basically tried to get as much out of the program as possible and came out with amazing resources and funding. However, I think I worked too much in my last year to the point of physical and mental exhaustion. I also spend 99% of my 14-hour workday now on AI (machine learning) and Robotics, and I mostly learned everything I know about building a startup from a book. So, I’d still do my time at Leeds, but I really wish I had done a part-time or full-time degree in AI and engineering. People fail to realize that most things in tech are relatively easy but time-consuming.

“Would you recommend that other entrepreneurs attend CU or college in general?”

I cannot emphasize this enough. Anyone that is already working on a startup and wants the right environment to scale, go to CU. However, the program is the most useful when you’re actually working on a venture. The programs are built around supporting you and your startup along the way, but if you don’t have something you’re working on, there’s no point to the degree.

Let’s get deeper into your startup, CLD-9. What is it that you do, and why is it important?

After spending a year heavily researching the supplement industry and interviewing hundreds of people on their issues with supplementation, I realized that the three main issues most supplement users are running into are:

• 1. The insane amount of research and testing it takes to find something that works for you. The population in America isn’t supposed to be spending hundreds of hours on research like a Ph.D. in physiology. It’s simply inefficient and built against the user. If they don’t want to spend the time researching, they listen to low quality sources of information that lie about the science or get something off the shelf that is usually far from ideal for them.
  2. If you do find a formulation of ingredients that works for you, you are most likely to spend $70-180 a month, which can get really expensive.
  3. Those who do opt to bootstrap their own personalized formulation have to take an insane amount of pills for their use case. Which, from my research, makes people slowly move away from using supplementation at all.

At CLD-9, we know that the solution is to make this entire process as quick and easy as possible. We have users simply talk to our AI that’s trained on a massive amount of research on supplementation, build a formulation that’s hyper-personalized to the mg, and get it shipped to you to drink. It sounds simple, but there are 2 main things we really innovated on.

• When me and my team first started on the AI, we wanted to just use what we thought was best, which was just an OpenAI model with some simple text scraping. Unfortunately, LLMs are not built to properly output information; they are text prediction systems that simply predict what word is supposed to be after the previous one based on their training data. When you’re dealing with people’s health, you can’t have a hallucinating AI, as you wouldn’t go to a hallucinating doctor for your problems. Thus, we built a 25 model system that ONLY uses our own data repository for recommending and educating our users on supplementation. Over time we intend to provide access to our AI model to democratize information and provide recommendations on basic health problems. For now, it’s making the most scientifically accurate supplement recommendations in the world.
• There are no machines out there that can personalize powder down to the milligram for different people and produce those differences on a mass scale. So, we engineered the first machine in the world that can do just that. Imagine made-to-order supplements that are specifically crafted for you. Right now, we’re trying to automate the process entirely using AI robotics.

That’s about it. We’re launching in early January of 2024 (finally) at CLD-Nine.com

“In your journey as an entrepreneur, what’s been the biggest personal challenge you’ve had to overcome, or are still working to overcome?”

Probably mental health. Spending 14 hours a day in a warehouse day after day for months isn’t that healthy. There’s always a deep sense of insecurity that something might not work, and things generally don’t work the first time. It’s up to the founder to keep things going and keep everyone else motivated while still having the stress of raising money and keeping the baby alive.

What do you think is the future of AI, and is there an AI bubble?

The VC community jumped onto AI without doing their research, just like web3. However, the best investors know what bubble-proof companies look like. Just because AI is really cool right now doesn’t mean it’s applicable to a certain problem. Do we really need more generative AI companies, or do we need more companies that solve problems in the most efficient way? It just so happened to be AI and robotics integrated for us, not because we wanted it to be, but because we had no other options and it was necessary to innovate. Bubble-proof companies are doing research to solve a problem that actually exists, and they will continue to provide the best solution whether OpenAI (the creators of Chat GPT) releases another dev day or not.

What’s in the future? Everyone currently thinks we’re on the brink of AGI*, but I’m not so sure. LLMs** are still pretty rudimentary and limited. They seem amazing at first glance because they are so good at replicating and predicting things, but real consciousness is not going to come from more data. The future is in replicating the architecture of our brain in machines to understand consciousness more. That’s what CLD-9 is working on in our 10-15 year timeline.

Editors Comment: *AGI = An artificial general intelligence is a hypothetical type of intelligent agent. If realized, an AGI could theoretically learn to accomplish any intellectual task that human beings or animals can perform (source: Wikipedia). **LLM = A large language model (LLM) is a specialized type of artificial intelligence (AI) that has been trained on vast amounts of text to understand existing content and generate original content (like Chat GPT).

“What’s the end goal for your venture? Where do you see the finish line?”

The finish line is, and has always been, giving me and my employees the resources to do what we love. If that’s through an acquisition or IPO, so be it. Externally, it’s to create a perfect symbiosis between people and the products they use. Right now, it’s supplementation, and it will be for a while, but it’s going to expand with more research and data.

“If you had unlimited funding and unlimited time, what would you do in the world?”

I would scale CLD-9 very quickly. We’re actually a really economically profitable company since we do almost everything in-house. After we scale and make the market more efficient, I’d like to spend a large amount of funds on building neurobiology and brain research labs that are directly connected with a massive AI research team to find out how consciousness really works. I have amazing talent around me in AI, and I think if we use my team’s architecture and research after it’s fully developed and have the funding to develop advanced neural systems similar to the brain, we can change the world significantly.

If YOU Are An Entrepreneur at CU, I’d Love To Tell Your Story

My name is Bennett Black and I am a columnist on AboutBoulder who tells the fascinating stories of entrepreneurs at CU. I am an entrepreneur myself and operate Bold Slate, a digital marketing agency for home service businesses. If you’re looking to be featured in one of my next posts, you can DM me on instagram @bennettblack.io

Bennett Black

Bennett Black is a CU alumni and the founder and CEO of Bold Slate. His unique style of journalism aims to tell the real, unadulterated stories of CU's brightest entrepreneurs. Join weekly on Wednesdays at 11am to read new CU entrepreneur stories on AboutBoulder.com.