A cup of innovation: Chatting with startup founders at their favorite cafés

Technology moves fast, and so does life in San Francisco. It’s all fueled by coffee—passion, and lots of coffee.

“A lot of founders live on coffee. They live on caffeine,” said Steven Songqi Pu, a Stanford computer science student focusing on artificial intelligence (AI).

Steven also makes movies, and this summer he decided to track down and interview startup founders at coffee shops around the Bay Area about how they use Amazon Web Services (AWS) AI tools to power their businesses. (Plus, learn what type of caffeine they choose to fuel them). Here’s what he learned:

Toby

Right away, Steven faced his biggest challenge: how do you translate the technical aspects of software development using AWS to an audience that might not be technical experts? His solution: focus on companies solving problems he could relate to—for instance, with translation.

Steven connected with the founders of Toby, a software which provides real-time, AI-generated spoken translation for video calls. Both founders grew up outside the US — Lucas is from Spain and Vincent’s parents are French—and between them, they speak seven languages (French, Mandarin, and English for both, plus Spanish for Lucas and German, Hebrew, and Esperanto for Vincent)[1] , so they are well aware of the value of being able to communicate in today’s hyper-connected, globalized world. Their vision is to enable a future where language is no longer a barrier to collaborating across national boundaries.

“Currently there’s a drastic pay gap between people who work in English versus those who can’t or don’t,” Lucas told Steven. “We want to remove that pay gap and enable anyone to work in any language.”

The pair used the AWS SageMaker machine learning (ML) model registry to train and deploy its language models across different regions, and the team also built their own load-balancer on AWS to help international users who might be having network traffic problems.

For Steven, their story was an example of how gen AI empowers founders to solve bold problems. “We've dreamt of a universal translator for hundreds of years,” Steven said, “and developments in AI now give teams like Toby the engineering platform to make it happen.”

How do you take your coffee?: Vincent’s go-to is a latte; Lucas opts for espresso.

Check out the episode featuring Toby on Youtube here.

Aragon

One reason startup founders drink so much coffee: they work a lot. A lot of that work is simply trying ideas over and over until, finally, something sticks.

That’s the story behind Aragon, a software which enables users to upload casual, candid photos of themselves and receive polished, professional headshots, generated with the help of AI. Wesley, Aragon’s founder, told Steven that he was down to his last $1,000 when he finally hit on the right strategy for his company.

“After analyzing user behavior, it turns out that people really wanted to generate photos of themselves. But not just any photo—professional headshots,” Wesley said.

With AWS S3, AWS EC2, and AWS RDS cloud computing tools, plus SageMaker for machine learning, Wesley and his team were able to iterate quickly and fine-tune their image-making models until they found a use case that clicked with customers.

“Gen AI is a really useful technology,” Steven said. “But I think a lot of people are still exploring a bunch of use cases and finding out what people love to use.”

How do you take your coffee? Wesley likes lattes.

Check out the episode featuring Aragon on Youtube here.

Switchboard

It took Switchboard co-founders Anupam and Cam about a dozen tries before they hit on their big idea. “We built a lot of fun toys and gadgets,” Cam told Steven. “But we never built something someone was willing to pay for.”

Eventually they asked their friends and family what they would pay for, and the answer came back resoundingly: an assistant. That started the duo thinking and eventually led them to Switchboard, a software which automates basic tasks like information-gathering and synthesis into a smart spreadsheet tool. The program relies on Anthropic's Claude through AWS Bedrock to extract data directly from the internet and organize it into meaningful insights.

One thing Steven found talking to all these founders is that having tools like Bedrock at their fingertips was the key to helping them grow. “They want to be able to scale very quickly,” Steven said. With AWS, it’s easy to add on and adapt their infrastructure as their needs change.

How do you take your coffee? Cam and Anupam both love cold brew with oat milk.

Check out the episode featuring Switchboard on Youtube here.

Evolution Devices

Most of the time, when we’re talking about AI-based tools, we’re not talking about the kind of tools you hold in your hand. But Pier and Juan, creators of Evolution Devices, are different. They’re building a machine learning-powered wearable that can help mobility-impaired people walk again—and they’re using AWS database and machine learning products to do it.

“We were inspired to build Evolution Devices initially as a solution for my dad, who’s living with multiple sclerosis,” Pier told Steven. MS attacks nerve cells, disrupting the systems that let your brain tell your body what to do. As MS progresses, even basic tasks like walking become a serious challenge.

The device Pier and Juan developed is called the EvoWalk. No bigger than an old-school pager, it straps onto the leg below the knee and helps the muscles fire at precise times to smooth out a halting gait and prevent fatigue.

As with any company that handles patient data, Evolution Devices has to abide by medical privacy laws. Pier and Juan use AWS EC2 instances so that both the patients themselves and their physicians can access and analyze the data from their individual devices. They also use SageMaker to train AI models and analyze data to track users’ progress.

How do you take your coffee? Pier is another latte-drinker, while Juan prefers matcha.     

Check out the episode featuring Evolution Devices on Youtube here.

ViralMoment

Getting in on the ground floor of a trend can be incredibly lucrative—if you can manage it. Historically, other than being incredibly lucky, there was no real way for entrepreneurs to measure and analyze where consumers would go before they got there.

Then came ViralMoment. Chelsie, the company’s founder, explained to Steven that the Amazon Rekognition system helps her company break down and analyze key elements from social media posts and start to draw out meaning.

“The power of ViralMoment is when we start to look at millions of videos across socials every day,” Chelsie said. “Then we can start to understand what’s trending, what’s going viral, what are people saying, so that we can measure culture effectively and at scale.”

That speaks to what makes AI such a powerful tool for startups, Steven said. “At the end of the day, it's all about, how do you create the most amount of impact with your limited resources?” That requires strategic thinking, intuition, and creativity, plus “a lot of grit and grind,” Steven added. Automating insights with AI powered by AWS is key to helping founders maximize their time—and their potential for success.

How do you take your coffee? Chelsie loves a flat white.

Check out the episode featuring ViralMoment on Youtube here.

Coffee powers founders, AWS powers startups

The best ideas were inspired by family, experiences, or research—innovation is counting on you for the next best thing. Grab a cup of coffee and visit AWS Startups learn how AWS' robust machine learning, cloud computing, and Gen AI tools will be your next source of inspiration.