Why AI Video Avatars Serve Some of the World's Largest Companies

It was a choreographed experience and made me feel as socially awkward as I have ever felt. I've accessed my laptop's camera to the avatar platform of Synthesia, a London-based AI video company. In about two minutes, you can read the lines loudly as you scroll through the screen.

Reconnect with a pleasant feeling before you move on.

Implement the following positive statement and my voice has energy. My tone is attractive and bright.

She laughs every time the wind tickles her nose.

I then followed with a series of positive and corporate-friendly statements: Imagine the impact we can make with this innovative idea!

The atmosphere wasn't bad, but it's odd. This process made me feel vulnerable, as if giving a part of myself to the computer. The whole process also took about 10 minutes, initially clicking, knocking on a coffee cup, and reading the script twice to improve some of my awkwardness from the first go-round. I submitted the video and Synthesia told me I have an avatar in one day. I think the short gestation period of my digital version was taking me to make the version that took me now 32 years.

Last year, 150,000 Synthesia users created AI-generated avatars. (Customers often choose from the company's 250 stock avatars.) And while I have built my avatars in the name of journalism, the most common use cases for Synthesia are business training and internal communication videos. Synthesia avatars are making impressive progress throughout the business world. The company currently has over 65,000 customers and says it serves more than 70% of the Fortune 100.

It is clearly a non-Hollywood form of video success. Enterprise customers can access Synthesia's platform and create videos using a single tool, just like they would pay PowerPoint's Microsoft. The appeal is the combination of cost and scale. If you have a large multinational company, making videos in multiple languages ​​is a costly, time-based effort that is historically limited to teams with substantial budgets. And most people absorb information more easily through video than through text. This is double to the burgeoning ranks of Gen Z workers who grew up on Tiktok and Instagram reels. To communicate effectively with these workers, managers need to do so on video.

“I think this applies to almost all transformative technologies. I think the real power of this is that new groups of people can do what they can't do before,” says Victor Riparbelli, CEO and co-founder of Synthesia. “What we've learned is that billions of people around the world aren't making the videos that they want to make today.”

Make digital “people” more realistic

Synthesia's technology is clearly a by-product of the generated AI waves, and features a large-scale AI model that is a training to understand both what we say and how to say it. This creates an avatar that mimics speech with persuasive realism. This is not through AI, which can predict movements and facial expressions, but through AI. The result was a rather naturalistic digital person. That naturalism, in turn, helps viewers and listeners feel more comfortable, but the video sometimes stops at the edge of a creepy valley.

Synthesia was launched in 2017, six months before ChatGpt created AI Boom. Riparbelli won Agapito through the academic web of Riparbelli, Matthias Niesner and Lourdes Agapito from a small Danish town who built the website as a child for a local store.

In the early days, the founders weren't focused on video generation yet, so instead they used AI to dub existing videos and focused their energy on floating as much as possible. Having struggled to raise funds at particularly important moments, they found Mark Cuba's email and sent him a cold pitch. Cuban responded within six minutes, triggering a 12-hour email exchange that went until 4am in the UK. (Cuba declined to comment on this story.)

“The first three or four years weren't successful,” Riparbelli said. “It was impossible to get funding. Technology didn't work. We didn't know exactly what it would help. It basically took us through the end of 2020 until the moment we hit the inflection.”

Synthesias' move to corporate video took place slowly, and then the by-product of talking to thousands of potential customers who wanted something better than PDF or PowerPoint at once.

“What we found was that many people in the work of a company that trains, marketing and customer support said, “I know I know I have an important message, and I know I haven't read my documents,” Riparbelli said. “They say: “I want to make videos, but I can't ignore making videos.”

Today, Synthesia's Fortune 500 and Fortune Global 500 customers all use technology in a deep, concrete and personalized way to their business, but they reverberate with each other. Merck Kgaa, a pharmaceutical company from Darmstadt, Germany, uses Synthesia to replace time-packed live recordings on product updates and use them for multilingual training. “We are thinking about great possibilities for avatars to make information more digestible and accessible,” says Florian Metz, global head of analytics and AI product portfolio at Merck KGAA. luck By email.

Around the world in California, ServiceNow uses technology in its global learning programs. At its sales onboarding academy, Pasquale Fontanetta, VP of Learning Solutions Studio, said 20 videos “reduced production time by 50%” and “enabled localization with a cost savings of up to $5,500.” (ServiceNow is the sponsor of this digital magazine edition.)

Another software company, the $360 billion German giant SAP, uses Synthesia videos throughout its sales and marketing process. “We see Synthesia as a communications platform, not just a training tool,” writes Andrew Steane, of the Business Management Office at SAP North America.

For Mondelēz International, owner of the Great Flood of snack food brands such as Oreo, Cadbury, Ritz and Sour Patch Kids, synthesia represents a solution to the problem of pile size in PDFs.

“If you said you were going to send us a 3-page long help article PDF with screenshots, would you read it?” said Geoffrey Wright, global solutions owner of Generated AI and Digital Experiences. If you can't say, he's cynical. “I voted people internally. What is the possibility that they're reading the article I sent to them, if it's not critical of the work, I'm critical of them, I'm critical of the obligation, I'm critical of what I was doing.

This year alone, Mondelēz has created 30,000 videos on Synthesia, Wright said.

The pitfalls of digital avatars

Synthesia has a shared iconic issues and controversy among those facing all generative video companies. For example, in 2023, the technology was used by a Venezuelan client to create state propaganda videos, and avatars were generated to mimic newscasters in the West. This episode crystallized concerns over political disinformation. Synthesia banned its customers, and has since strengthened its policy and moderation system on news and political content, and has continued to do so for years.

The company is part of a broader discourse about AI and potential unemployment. For example, does more synthesia videos mean less work for real professional video producers? And, as with all generation media companies, Synthesia has drawn concerns from actors about how they can be used when working on a platform. This year, Synthesia has taken steps to compensate actors who work beyond cash, and announced a $1 million equity fund that will provide the company's stock to actors who work to create AI avatars. These actors retain direct economic benefits for Synthesia as the company grows, and the company says the program will help build long-term dialogue with the actors, just like the others.

Synthesia works with what Riparbelli calls “Three CS.” It adds that this is an agreement, control, and collaboration, and that you will not create someone's avatar without express consent. “There are other players in this field who make fun videos of celebrities to get viral moments rather than misinformation purposes,” he said. “It's something we decided, for example, to never do. We don't revive the dead people. We draw a line. If that person cannot give explicit consent, we don't do that. So when you make an avatar on this platform, you can't upload footage of someone you find on the internet. It has to be you.”

The company also seriously manages the types of content permitted on its site and actively collaborates with governments and regulators. For customers that are not identifiable companies, Synthesia turns away.

“We say no to any business,” Riparveli said. “If you're working with big companies, they don't want to work with small guerrilla companies. One day, the next day they're selling big enterprise contracts, they're waving your CEO deeply into a viral moment.”

Mondelēz's Wright evaluates the entire market for generated video products every six months, pointing out that Synthesia has serious competition with venture-backed competitors Heygen, Colossyan and Hour One (which WIX recently acquired). However, despite this semi-conflict process of reassessment, Mondelēz has been using Synthesia for the past three years. why?

“They're the best for businesses,” Wright says. Because Synthesia has its own clever handle on the security and data protocols that the biggest companies need.

Approximately 24 hours after the recording session, I reopened the Synthesia website to see the avatar. Virtual Me is not as uncomfortable as I thought it would be. It's just as weird as watching yourself on a video call. This digital version of me has some microexpressions and a smooth voice that undoubtedly derived from mine.

“Hello, I'm your synthesis avatar,” my new creation told me. “Did you know I speak many other languages?”