"The future is here, it just isn't evenly distributed"

The best startups are started in garages, which aren't particularly known to be full of investors. On the other hand, the investor's challenge lies in finding these startups before they grow too large, and the question then becomes how to balance these two situations.

One the biggest challenges of our ever-growing world is the disconnect between investors and small startups that have valuable solutions. Because technology is such a crowded marketplace, it is very difficult to systematically find the companies that most direct solve the issues that are the most pertinent to society. Therefore, at a very high level, our piece of software seeks to bridge this gap, identifying society's needs and then using them to search for companies that are best suited to solve these issues. This in turns can serve to predict which companies are likely to be the most successful by seeing which are best equipped to tackle the modern market.

In summary, we use cool tech to find cool tech.


Our software can be broken down into three main components: an web crawler that looks for issues, a web crawler that looks for solutions, and a final evaluation that joins the two.

The first web crawler trolls through social media posts in a given topic of interest, and uses the google natural language processing API to find complaints. The complaints are summarized down to keywords and sent over to a server which stores them into a database. When enough data has been collected, it is transformed into a massive graph and analyzed to find the most connected set of nodes, aka the topic of conversation.

For the second system, a second webcrawler searches for companies that fit the conversation. It does this by filtering companies out of a search, and then identifying the ones that seem to solve the problem presented earlier on.

For the final system, the algorithm breaks down the top companies and breaks down their service and mission into a few key terms. Then, we use unsupervised learning to analyze the similarity between the content on the website and the initial search terms from the weighted graphs. The top companies are then presented in ranked order to the user for them to look further into for themselves. Links to these companies are then reposted on the same social media forums where we initially found these grievances, and their popularity is automatically analyzed to gauge how well people will latch on to their service. This serves to validate our results with the very people we are finding a solution for.


For the topic of "bank", the first webcrawler returned the key topics "bank card", "investment" and other similarly related terms. The second web-crawler took these keys and returned 13 unique links. Of these, 8 links were related to various banking services located in India. This seemed strange, but some research uncovered that the Indian banking system is currently experiencing a major cash shortage, marking a fantastic opportunity for innovative companies to capture market share. The other services found included Monese (an online banking service), Verifi (an eCommerce verification platform), Grameen (a Microfinance platform for people in low-income countries), the very classic fintech startup known as PayPal and NASDAQ, a major American stock exchange.

What else?

While the software was designed to analyze technology trends, it was build in a modular fashion and has several powerful tools that can be repurposed. The first web crawler in particular was very good at analyzing cultural trends in various interests groups. For example, the main topic of discussion in BBC news Facebook headlines was mental health, while the main topic of discussion in the comments section was Brexit and the immigration crisis.

Share this project: