When Isoken Igbinedion was 10, she had a “very dangerous encounter” with chemical relaxers that caused her natural hair to fall out. She spent the next 20 years using extensions to give her hair a chance to grow back. And around that time, she realized how much friction there was in the market for hair products and services.
“In those 20 years, we’ve seen very little innovation used to improve the process of buying and installing hair for customers or the manufacturing process,” says Igbinedion, 30.
That’s how the vision for Parfait came to be, a direct-to-consumer platform leveraging AI to better customize wigs, from head size to lace front hue.
Co-founded by Igbinedion, along with her sister, Ifueko, Marlyse Reeves and Simone Kendle, Parfait is coming out of beta with a waiting list of over 10,000 customers. The company recently announced $5 million in a seed round led by Upfront Ventures and Serena Ventures that will help scale up production and automate its supply chain to better gain a foothold in the wigs and extensions. should grow nearly 15% to $13.2 billion by 2026.
“At a time when technology has begun to make life easier for ordinary people with driverless vehicles, smartphones, improving our ability to connect with people, little attention is paid to solving the problems of marginalized communities” , says Isoken, now CEO of the company. “And this issue is deeply felt by black women, especially in the beauty industry.”
The other problem, ultimately, was the very data Perfect was trying to use to serve its core demographic. While the team tried to account for factors such as skin tone, existing open source datasets for facial recognition were simply not inclusive enough.
“Even the algorithm for recognizing a face in an image is biased due to the nature of the dataset,” explains Ifueko Igbinedion, CTO of Parfait. “There’s not really a huge amount of black women represented.”
In recent years, the biases of AI have come under scrutiny, especially facial recognition software. Wrongful arrests and discriminatory recruitment illustrate how communities of color are routinely subject to the vagaries of technology presented as agnostic but with little regulation of how code is written — or if code even considers them at all. Organizations including Timnit Gebruof DAIR and Joy Buolamwini‘s Algorithmic Justice League is pushing for fairer AI in Big Tech, and Parfait aims to do the same in the beauty industry.
“A lot of times we think that AI uses aren’t really useful in our community, like our community doesn’t deserve this kind of technology,” Ifueko says. “But we’re really showing that not only does our community deserve it, but our community needs it.”
Perfect customers choose the texture and cut of their wig, then they are asked to take several selfies from different angles to get a good match between size and skin tone. From there, an in-house human stylist accesses the AI prediction before customizing the wig, such as plucking the hairline, tinting the lace front, etc.
“I think it’s hard to say that an AI will be perfect,” says Ifueko. “And so we really focus on that human interaction in the loop, so that we have an accurate prediction.”
In terms of price, Simone Kendle, co-founder and CMO of Parfait, notes that the quality of Parfait’s wigs would typically be above $2,000 but, because of its DTC and they have automated elements of production, the wigs cost between $400 and $800.
“We’re reducing that price dramatically and we’re also accelerating turnaround time,” she says. Instead of waiting months for a wig, Parfait’s turnaround time is five to seven business days.
Perfect strives to have viewing capabilities on its site, similar to how, for example, Warby Parker allows customers to virtually try on eyewear. The company also plans to adapt its manufacturing process.
“Manufacturing in this industry is so archaic,” Isoken says. “Our current vision is to create a fully automated manufacturing pipeline. We’ve already done feasibility testing that has shown us that we can use computer vision technology and robotics to start creating these [wigs] from the source. »
As Perfect’s datasets grow, there could be an opportunity to have a broader impact on the beauty industry.
“We can extend [our technology] to all other types of beauty or fashion concerns where existing datasets don’t obtain the characteristics of your lip, cheekbone or ear shape,” says Marlyse Reeves, co-founder and COO of Parfait. “We have found solutions to some of these issues, and I think we will continue to find solutions to some of these issues. And hopefully this can be an example that if you just change your way of thinking, apply more diversity to the data you use, solutions present themselves.
That said, the question of data security still stands, especially with something like buying wigs where customers historically haven’t needed AI. Knowing that Perfect can be a tough sell for some customers, Isoken doubles down on transparency and the message that this technology works for them, not against them.
“We use this data to create products and experiences for our community,” Isoken says. “There’s been this idea that technology is either too dangerous for black people, or it’s not made for black people, or that black people shouldn’t engage in technology.”
Of course, Perfect isn’t just for women of color. The company is launching a campaign illustrating the wide range of women who wear wigs and the wide range of reasons why. However, as Isoken notes, “we want to continue to use our data so that we can continue to create products that start by solving the problems of the most marginalized people, because internally we believe that if we solve for the margins, we solve for everyone”.