Seth Raphael has spent decades working in tech. The last 10 years he was at Google where he built the first version of Google Photos and then ran teams in prototyping, augmented and virtual reality, and productivity tools.
He also has five kids.
His latest project combines those two elements. Hello Wonder is designed to help children navigate the internet in a safe and healthy way while giving parents, like himself, an ability to not only see what their kids are interested in but set controls when needed.
“I built it because I needed it,” he said, adding that the years of the pandemic and Zoom school and screen time were making him the bad guy in his family when it came to setting limits.
Wonder is a kind of AI assistant for kids that helps them navigate the internet as an interactive user interface. Wonder can feed websites and videos to children based on queries or answers to questions Wonder asks the child.
It can also see what a child is doing and ask comprehension questions of the child or set reminders, like getting a homework assignment done.
Parents input information about the child. As the Wonder chatbot interacts with the child, it learns as well.
Parents also get text messages about the videos and sites the child has been visiting, which opens up opportunities for parents to engage with their kids on topics they are interested in, said Raphael. And caregivers can set controls.
For instance, parents can direct Wonder to exclude information about guns and redirect any questions about firearms to a potential topic of interest, like the science fireworks.
“It’s like Mary Poppins for the internet. Mary Poppins will help you discover the world but she will also help you clean your room,” he said.
Raphael, who studied at MIT Media Lab, left Google earlier this year to build Wonder. His co-founders also have deep industry experience. Brian Backus was most recently executive game producer at Amazon, and Daniel Shiplacoff was most recently design director at Google.
“We are all startup people who have done time at big companies,” Raphael said. “We know how to build the first version and get to a billion people.”
Raphael wrote his first neural network, which is a kind of AI that teaches computers to process data like the human brain, in 2003 while still an undergrad.
“I wrote a paper about the future of AI teaching kids,” he said. “It’s not creating smarter tools but tools to help us become smarter.”
Six families, including Raphael’s, are testing the system.
The team is four people and soon to be five, said Raphael. The company recently wrapped up a pre-seed round, which surpassed its target. Investors are from the Bay Area and include angels and some venture capital funds.
As other founders have noted this year — and the data has supported — it is a lot harder to raise money right now. Raphael said he had better than expected reception for his fundraise but it was still difficult as investor risk tolerance has dropped compared to past years.
At this point Wonder’s assistant is aimed at children between five and 12 years old. But, the plan is the assistant would be able to grow with its users.
