A Challenge, a Team, and 24 Hours.

- Alex Reibman, Excel Ventures 2016

It’s a humid Summer Thursday night in the heart of Tel Aviv, but not everyone is celebrating the end of a busy week. As the vibrant city nightlife hums in the distance, a high-pitched screech from my overheating laptop snaps me back into focus. I grind my bloodshot eyes to peek at the clock — It’s 3am. I’m sitting at a communal table in a WeWork in the Southern part of the city. The light chatter and incessant clicking of other programmers in the room does little to calm my stress-induced migraine as I start to readjust the hyper-parameters of my speech summarization algorithm. I take a look to my right and see my teammate Dominik Hempel rubbing his eyes in exhaustion after spending nearly an hour on a Node.js server error that refuses to go away. On my left, Marcel Aphraim lets out an exasperated sigh after taking a final sip from his 4th cup of coffee. Our bodies — and our minds — are beginning to waver.

In less than 12 hours, we will deliver a functional product to a panel of judges consisting of the professional investors, successful founders, and the CTO of Bank Hapoalim. Yet despite the Gordian knot in front of us, our team prevails and walks away $2,000 richer.

Dominik, Marcel, and I participated in a competition called Hackathon.AI.

A “hackathon” is a competition where small teams come together to build a solution to a problem in a very short time frame. Hackathons are generally supported by big name sponsors such as Microsoft, Google, and Amazon to showcase new technologies and encourage innovation within technical communities. With a large prize at stake and enough caffeine to fuel a small country, teams push themselves to their physical and mental limits in order to achieve an extraordinary amount of progress in a short amount of time. Although the effort can be exhausting, teams have the privilege of consulting professional mentors and industry experts to help guide them through the process.

OUR IDEA

Hackathon.AI was themed around building a product that implements AI, personal assistants, or chatbots. With this in mind, our team spent a lot of time debating and validating our ideas. Initially, we had a few ideas in mind:

  • A chatbot to help you trade and manage Bitcoin.
  • A phone keyboard that generates suggested responses to incoming text messages.
  • A camera interface for cash registers that measures customers’ sentiment at checkout.

If we didn’t know any better, we probably would have just jumped into one of these ideas without a second thought. But thanks to some of the lessons Dominik and I learned from Excel Ventures, we had some advantages over the competition in terms of strategic planning:

  1. Scratch your own itch.
    It’s tempting to think of how to solve a problem that someone else might have, but your secret weapon for creating a compelling, sellable product is creating a solution to a problem that you personally face. If your product can solve a problem that you struggle with, you’ve already sold your first customer — You! If you can identify one of your own pain points, then others likely face the same problem as you and would love to find a solution. Judges evaluate teams based on how useful your product is to a wider audience, so if you can demonstrate that your product solves a gap in the market, then you have a strong advantage over the competition.
  2. Incorporate feedback as soon as possible.
    Since hackathons are often sponsored by corporate partners, industry experts and mentors are present available to help. Consultants, lawyers, PR experts, computer scientists, and other highly skilled mentors are freely available to help your team at any step of the process. During our ideation process, we took advice from a marketing guru as well as one of the lead engineers from Viber to better understand the chatbot and personal assistant landscape. During the building phase, we took advice from a Natural Language Processing expert to develop an effective coding strategy. Mentors are often overlooked at hackathons, but their feedback is often critical for developing a winning solution.
  3. Agility beats stability.
    24 hours is not enough time to build something perfect. Use this to your advantage. It’s better to experiment and pivot with your ideas than to dedicate all of your time to an uncertain one. Seek the feedback of mentors, experts, and outsiders and incorporate the bits and pieces that improve the feasibility of your idea. A naïve decision to work on an untested idea will minimize your chances of winning. If you don’t dedicate a lot of time towards building something specific, then the cost of pivoting to a different idea will be low. This ability to change your focus at a moment’s notice gives your team a better chance of building a great idea.

After some serious debate and discussion, we ended up abandoning our initial ideas. None of us knew, held or traded any Bitcoin, we found the use cases for quick responses highly limited, and we had no idea if stores seriously considered tracking their customers’ feelings.

Instead, we took a different approach and tackled a problem we all personally experienced — getting distracted during business meetings. Business meetings are common in large companies, but attendees rarely retain much of the information. Many meetings have a dedicated note-taker who summarizes the information, but hiring someone just to take notes is far from money well-spent.

Given recent advancements in artificial intelligence, we found this to be a specific pain point that could be innovated upon. So, we created Summit: a meeting-summarizing personal assistant that records and summarizes your meetings.

HOW WE TESTED

After validating Summit with several of the mentors, we realized that we had a solid idea to work with. We utilized several of the tools and API keys generously provided to us by the hackathon and began working on our product. First, we developed a Node web application used the Microsoft Cognitive Services API to provide speech transcriptions. Next, we implemented a Natural Language Processing algorithm to condense the summarized text by ranking important sentences.

PRESENTING THE VALUE

24 sleepless hours later, and the announcement comes that presentations are ready to begin. We just fixed the last bug 30 seconds prior. Despite hours of nonstop work, we only have 5 minutes to show off everything. The key to winning, however, is not to spend the entire time demonstrating your solution. Rather, the bulk of your time should be spent explaining the problem and why your solution uniquely solves it. Ultimately, your demo is a proof of concept.

For our presentation, we gave evidence that supported our claim that business meetings were highly inefficient without written summaries or transcripts. This was a problem we had all personally faced, but had confidence in this hypothesis since we had devoted a considerable amount of time discussing the idea with mentors as well as performing research.

For the demonstration itself, we showed examples of our product compressing lengthy conversations into easily legible paragraphs. Furthermore, we took the audio recording of the demo as we spoke and showed off the summarized results.

WHAT WE LEARNED

Needless to say, the judges were impressed by our solution and awarded us with a $2,000 prize. Despite the physical and mental toll the hackathon took on us, we were confident in ourselves because we took a critical yet sophisticated approach in designing our product. We knew that by putting the extra effort into validating the idea, we would be sitting on a winning solution. Afterwards, we spoke with one of the judges, the CTO of Bank Hapoalim, and he told us that if this were designed for Hebrew business meetings, he would have bought it on the spot.

To put things succinctly — To build a winning idea, think before you leap, but leap in a lot of different places.

Photo Credit: Ronen Mendzitsky, Gadgety

Alex Reibman is a data scientist with a background in economics, cybersecurity, and innovation and has worked in Tel Aviv’s high-tech sector for over a year. Alex graduated from Emory University in 2016 with a degree in Economics and Philosophy and was an Excel Ventures Fellow in 2016.