AI, The Problem Solver

With its tagline “A lot can happen over coffee”, Café Coffee Day made an entire generation fall in love over several cups of the earthy beverage. Abhishek Sharma and Dipti Tandon, the founders of professional networking start-up Coffeemug.ai, are looking to tap into the same sentiment with a matchmaking algorithm.

Matchmaking, in itself, is not new to the founders. Tandon was also the founder of JeevanSathi.com before she sold it to Info Edge and went on to hold top-level positions at MagicBricks, Times Internet and Lenskart. Sharma started off as an entrepreneur, turned into an HR executive for Times Internet where he met Tandon and is back in the entrepreneur’s chair. He has also held top positions at Ixigo.com and DineOut but the idea of Coffeemug.ai, he says, lingered on since 2018.

“The chemistry is very different when you meet someone one-on-one in an informal setting as compared to when you meet someone for an interview,” says Sharma, adding that they wanted to help people meet professionally but over a cup of coffee.

That was the original plan until the pandemic struck. The Singapore-based start-up, which was founded in March 2020, had to rejig and become a virtual platform.

The platform, which targets entrepreneurs, investors and business leaders, has a gatekeeper algorithm. While one has to apply to join the platform and checks are done, it also invites members.

The profiles are scored based on credentials and profiles with complementary scores are matched. Meetings are set up and members are also asked to rate them. The only other way that members can connect with someone is by expressing an interest and the profiles are matched only if the other member is also interested.

Coffeemug.ai says that it differs vastly from veteran LinkedIn as it stands firmly by its novelty of filtering. Unlike the networking giant, the start-up is not a social networking platform where members can post images, videos and status updates. The platform also boasts of privacy wherein members can connect only when the algorithm matches them. For now, it plans to focus on three buckets – job, investment and mentorship.

But if the platform is free, where does the money come from? “We’re focusing on helping companies look for candidates for their higher-level positions like CFO (chief financial officer) and CTO (chief technical officer),” says Sharma. Coffeemug.ai charges some percentage of the selected candidates’ CTC. So far, it has helped out about 15-20 internet companies, including Lenskart and Haptik. Coffeemug.ai recently raised $625,000 from US-based Paradigm Shift Capital and AngelList India in a pre-seed round and is looking to make about $1 million dollars in FY22.

Currently, with members from 10 different countries on its platform, the start-up has a network of 50,000-60,000 members and aims to touch 100,000 by the end of 2021.

Over the past few years, visuals of politicians and businessmen making scandalous statements have been circulated across the internet, most of which are far from reality. The fact is that while the visuals may be original, the voices are manipulated.

Kroop.ai hopes to make finding these manipulations at the click of a button.

“During election time in the US, there were a lot of these popular deepfakes doing the rounds and I realised that there wasn’t any specific tool available that could validate their authenticity. There are a lot of fake news busters but those are text-based,” says founder Jyoti Joshi, a Melbourne-based researcher.

Deepfakes, Joshi explains, are synthetic media generated by artificial intelligence where someone is made to say and do something through manipulation of facial expressions and audio.

Joshi and IIT-Ropar alumni Milan Chaudhari and Sarthak Gupta, who were already working in the tech space, formally established Kroop.ai in February 2021.

“The problem is very niche and, in the Asia Pacific region, there are no companies (dealing with it). The threat is new and so is the whole set-up around its detection,” says Joshi.

Kroop.ai has an API, more suited for large-scale systems that can integrate it into their existing system for verification. It also has a website where a piece of media can be uploaded. There are two more access points to the tool—the mobile app and the web scraper, a keyword-based system that scans the medium for potential manipulation.

While the website just provides the probability of the video being manipulated, the tool offers a detailed analysis in terms of the percentage of manipulation and the aspects of the video that have been manipulated. On an average, it costs a user $1 per minute of video analysis.

The founders say that their current focus is on the B2B market and they are currently in talks with two fact-checking companies, one US-based cybersecurity firm and the other a leading private bank in India. The company, which received its first set of funding from 100x.vc in May 2021, has seen a demand for the tool in the BFSI sector with a growing number of video KYCs.