The Croatian entrepreneur using facial recognition to tell how you really feel

Josipa Majic speaks to Hazel Sheffield about founding Tacit, the biometrics data company that maps your hidden emotions

Josipa Majic is the founder and chief executive of Tacit, a startup that uses biometric data to measure emotional responses

Josipa Majic, the Croatian entrepreneur behind biometrics startup Tacit, has not always found it easy to explain her career choice to her family.

“For my parents, it was like, ‘You’re never going to get married’,” she says. “Croatia was a part of Yugoslavia. so there were no real companies – we had socialism. It was difficult to explain to them, because the local perception is that entrepreneurs are tycoons and connected to the mafia – not something you would want your daughter to be.”

Croatia joined the EU in 2013. Its ascension boosted the county’s exports from below 40 per cent of GDP in 2008 to above 50 per cent in 2018, according to the World Bank. But tech entrepreneurs from the country have faced challenges including lack of public funding and education to support the sector.

“In 2019, [entrepreneurship] sounds more logical as a career path [in Croatia] but, a couple of years ago, with two women, no family support financially, it was quite unusual,” she says.

Tacit was founded by Majic out of the ashes of another biometrics research company called ID Guardian. It has since expanded to a team of five working with other specialists in Boston in the US and Singapore to provide insight for businesses based on analysing a respondent’s emotions, from their heart rate to their eye movements, to tell how they truly feel rather than what they are trying to portray.

So far, Tacit has been used by the gaming industry to create a performance algorithm for players of video games, allowing them to get granular insights on their game, and in-product development, when analysts might be trying to judge whether a customer prefers one version of a recipe over another.

“People have a hard time articulating how they feel about taste and smell,” Majic says. “We can map out, millisecond by millisecond, how someone is feeling by using their biometric data.”

Tacit captures data across 11 data points, using two different products. Tacit Light gathers data remotely, using the camera on a smartphone or a web-camera. Tacit Pro includes hardware that has to be worn by the respondent.

Majic says the company uses the light versions of the software to make hypotheses that can be proved using the professional product, which offers more conclusive data. The data collected includes facial coding, speech analysis, eye tracking, skin temperature, heart rate and the amount of oxygen in the blood.

On the Tacit website, videos demonstrate a dizzying number of potential applications. In one, a user’s biometrics are tracked while they search for teeth-whitening products, showing they are excited when they start to research the topic, disgusted when they discover a home-made whitening video, satisfied during scrolling and the moment that they decide to buy a product. In another, the emotional response of someone is tracked while they are looking for baby food, showing dissatisfaction with high prices and disappointment with ingredients.

Perhaps even more exciting is the potential for using biometric data to track emotional responses during job interviews and shareholder calls. Another video shows Tacit software analysing the tone and pitch of Elon Musk’s voice as he answers questions on a call about the company’s financial results. The video suggests at what points Musk’s speech indicates stress, particularly around details about the Tesla Model 3.

The local perception is that entrepreneurs are tycoons and connected to the mafia – not something you would want your daughter to be

“You can give the perfect answer but, if you put it through a speech sentiment analysis, it might not be in sync with what the person is saying,” Majic says. “There are moments when it gets interesting on earnings calls and recruitment where someone could say they are really good at something, but the results show something different.”

Tacit has developed the algorithms by extracting metadata from white papers and then testing it in labs. It is currently working with researchers at the University of Zagreb on a project quantifying the levels of empathy in convicted criminals. “One of the things they are doing is showing virtual reality content of emotionally charged scenes to provoke a negative reaction, and seeing to what degree the criminals react,” Majic says.

Is there any assignment that Tacit would refuse on ethical grounds? “We would say no to anything that is a ‘Black Mirror’ moment,” Majic says. “For example, when the person doesn’t know data is being collected. We don’t want to be in the position where someone down the line uses that dataset for something it was not intended for. Anything that sounds too vague, or requires sharing raw data, is unacceptable to us.”

Under EU law, citizens are protected from having their biometric information linked with their identity and shared with third parties without their consent. Under GDPR, companies collecting biometric data also have to boost the security around that data or face fines of up to €20m. Nonetheless, anything that can be measured remotely, such as speech analysis and facial coding, can be captured without consent, Majic says.

“The EU is good at raising the level of awareness on data protection,” Majic says. “But Asian companies are prolific on the scale they deploy facial recognition, facial coding and everything that can be captured remotely. We’re all aware that whoever has the biggest data set is winning – and the EU will not be winning.”

Tacit has not received any investment, preferring to stay private and sell its services rather than seeking outside funding. Majic says this decision was influenced in part by her early experience in trying to build a biometric teddy bear that acted like a Fitbit for kids. That project attracted a small amount of funding, she says, but nowhere near the amount needed to produce and export the hardware.

Majic says she was “soul-crushed” when they had to fold that business. “I was 23 at the time, so I lacked some hands-on experience, but we realised that we have built a lot of expertise in sensors and we could do stuff that was more realistic, analysing biometric data rather than manufacturing hardware.”

Majic’s pragmatism comes from her experience of growing up in a country where her options were limited. “The rationale was quite simple,” she says of her decision to become an entrepreneur when she graduated in 2013, at a time when the Croatian economy was still suffering from the after-effects of the financial crisis. “Me and my co-founder realised that there would be no dream job, so we decided to build something on our own.”

And have her parents finally come around to the idea? “The perception [of entrepreneurs] is slightly changing now, because of a few [from Croatia] who are amazing,” she says. “But to this day, we still have interesting debates about whether it’s worth it, because for them it was never an option.”