When I evaluate startups, one of the most important variables that I look at is the problem the startup intends to solve. How deeply does the startup understand the problem? Vibes aside, do they have any real evidence that people will pay real money to solve it? How well are the startup’s reps able to articulate the negative drag of the problem to people who aren’t focusing on it day in and day out?
In Challenge #2, 42 startups participated in this round, and 20 passed this test. For the startups that didn’t move on in this round, the main issue is either 1) you never detailed the problem, instead jumping right ahead to the solution, or 2) the description of the problem was confusing, unbelievable, or both.
The 20 startups that passed the test all clearly articulated the problem they solve in terms that non-experts can understand. They illustrated the scope of the problem, and they gave evidence for why the problem they target will continue to get worse if it’s not addressed.
AI-powered problem solving
| Startup | The problem they solve | Why it matters |
| FeatureByte | Anyone, anywhere in the world with a browser can have a conversation with AI or create prose and art for free. But the story of AI in the enterprise is remarkably different. Technologies like AutoML and MLOps have democratized ML modeling, but the process of preparing data, deploying pipelines, and managing it at scale is where enterprise AI breaks down. | AI data management (or feature engineering and management) is laborious and expensive and requires deep domain, data science, and data engineering expertise. Unless the process is automated and democratized, the promise of AI everywhere in enterprises will remain just that, a promise. |
| Sailes, Inc. | Many professional workers spend their time on repetitive manual tasks. In the case of sales executives, sales enablement tools force reps to live in a hamster wheel of mundane tasks aided by small automations such as reminders and calendar software. From contact databases to marketing automation, a web of orchestrated software creates an obstacle course around sales executives. | According to Accel, most sales teams are now using 12 different point solutions to close a deal. This status quo is a problem because most salespeople spend more time searching than selling. As AI advances, sales reps who still perform tasks manually will lose opportunities to reps using modern tools. |
| Selector AI | Network operations teams face numerous challenges when triaging incidents and resolving them. As networks sprawl, teams need better tools to gather heterogeneous data and then correlate data to form insights and identify root causes. | During an incident, operations teams are often challenged to identify the root cause of a given issue. These challenges are magnified by numerous point solutions that tend to present only small pieces of the overall problem. |
| SimplicityDX | Nearly 90% of US online shoppers complain about poor experiences when clicking through from social media to a brand’s e-commerce site. These poor experiences result in very high bounce rates (up to 90%) and very low conversion rates (0.5 – 1%). | The root of the problem is that social traffic is almost always sent to d brand’s e-commerce Product Detail Page. These pages were never designed for traffic from the edge, but the alternative of trying to build dedicated landing pages for every social post just isn’t practical. |
