top of page

Round Table Discussion Highlights 1: Enterprise AI - Startup's"Last Mile Challenge"


This article features the discussion highlights from 1st Fellows Fund Round Table Discussion took place on July 17th, 2021. Speakers include: Vijay K Narayanan, Daniel Kokotov, Lei Yang, Haixun Wang, Xuedong Huang, Anshul Pande, Gang Hua, Alex Ren


Senior fellows of Fellows Fund at round table discussion
 

Full video of Fellows Fund Summit 2021: Tech Entrepreneurship in the AI Era

Fireside chat between Eric Yuan and Xuedong Huang

 

Enterprise AI - Startup's "Last Mile Challenge"


Alex Ren

I was talking to Gang last night regarding Enterprise AI’s ecosystem is quite fragmented in terms of data, application, etc. Where do you think the opportunities for startups lie?


Vijay K Narayanan

In the last 20 years, I've spent almost 10 years in the consumer space and 10 years in the enterprise space, applying machine learning technology. AI development has disproportionately been applied to the consumer space, however, there remain many unique problems and challenges in the enterprise space. I usually call it “the last mile problem”, meaning applying AI in the context of a process or embedding AI into a system.

SVP Engineering at ServiceNow Vijay K Narayanan

The key is how do you translate the outputs you get from a recommendation or prediction into next step execution? In the consumer space, it's fairly easy. Think about Twitter, Facebook, LinkedIn, or even Pinterest, you could actually take the predictions and put together an end-to-end system. When you deliver it in enterprises, this “last mile problem” happens to be very different. How do you solve the same problem but at scale? Each individual customer becomes an important challenge. So I think that is a good opportunity for enterprises and for startups. I think that's where the value gets generated, ultimately.

Xuedong Huang

There are two ways for startups to succeed. You can create a breakthrough and outperform platforms like what Microsoft, Google and Amazon did. That's great, but unlikely to happen for startups. Or you can just create a new stack from beginning to end and you are more likely to own a booty call. An even better solution is you take Microsoft Azure AI to customize, leverage that to stand on the shoulder of Agile AI. The net return on investment could be tremendous.

Gang Hua

I think XD brought up some really good points. That's why I'm working on a vertical in a startup. By developing everything by ourselves, we can integrate it into the solution. I think what Vijay mentioned about the “last mile problem”, I read it from two aspects. The first is that you cannot have a general model trained on general data to work on every single scenario in the enterprise space. The other is you really have to shape your learning mechanism - the AI capability - to be fast enough and enable it to be used by low-cost engineers instead of high-cost machining learning engineers or deep learning experts.


Alex Ren

Do you think, enterprises should invest more in the front-end such as customer experiences, or the back-end such as operational efficiency?

Vijay K Narayanan

Both. In the backend, tremendous, tremendous opportunities for automation. See where ServiceNow is at. But at the end of the day, back-end is still serving somebody in the front-end. Maybe it's not an external customer, maybe it's only your internal employees, how do you handle the front-end experience?

Xuedong Huang

Think about Operation System in general, whether it's iOS, Windows or Android, not many startups created a competing OS. But they're probably 100,000 times more startups working on that ecosystem. They're not doing the whole stack. AI is at an early stage, but you can see the horizontal layer is happening. My suggestion is to leverage what’s available and improve the efficiency.

Lei Yang

It's an interesting question to ask, where are the startup opportunities in the enterprise AI space. A lot of our discussions have been fantastic, but they're abstract. I was trying to get on to the tactical problems that we encounter in managing big companies. I think there's a lot of opportunities in using AI to solve the efficiency problem.



VP Engineer at Twitter Lei Yang

In the consumer space, machine learning and AI is to optimize the self-efficiency problem. In the enterprise world, in terms of solving business problems, there’re are a lot of opportunities as well. For instance, do you feel like your recruiting effort is good enough? I was talking to my manager yesterday about how hard it has been to recruit suitable people. It’s a time-consuming process. But where can I get help with that? I've known several companies trying to use machine learning to do matching and suggestions, but that’s not good enough. I'm just giving an example of how many opportunities for startups to work in the enterprise world.



#ai #MicrosoftAzureAI #enterpriseai #startupopportunities #machinelearning #aiapplication


 

Join Fellows Fund club to receive more updates about AI Startup Investment Opportunities



372 views0 comments
bottom of page