impress.ai is a deep technology startup founded by Sudhanshu Ahuja (CEO), Vaisagh Viswanathan (CTO) and Amrith Dhananjayan (COO). The idea is to provide recruiters with an AI-powered platform to help evaluate candidates. The result? A ten-fold improvement in hiring time and efficiency. Currently, the service is offered to enterprises, startups and individual recruiters who can use it to build a customised chatbot-based job application. The embedded machine learning algorithms adapt to the organizations’ preferences over time, thereby simplifying their work. Started only in late 2016, Impress has already managed to secure an undisclosed amount of funding from the Javelin Startup-O Victory Fund to develop their virtual HR assistant further.
In this interview, Vaisagh talks about the inception of and technology behind impress.ai, what it means to be a tech entrepreneur, and more. Here is an excerpt from our conversation. Served with a shot of Espresso, of course!
Let’s start at the very beginning. You have been working with Sudhanshu and Amrith long before Impress AI was formed. How did you meet them?
I was doing my PhD and Sudhanshu started this company called Ideatory- as in Idea Factory- in 2013, which was about bringing R&D services to SMEs. So he contacted me and I started working with him part-time. I liked the idea and we kept meeting up. He reached out to Amrith too, who also started working part-time. Eventually, we became actively involved to the point of being co-founders.
So you started off with a completely different product in a completely different industry. How and when did you think of entering the recruitment space?
We pivoted a lot during the first year or so. As I mentioned earlier, we started off with Ideatory- a R&D service-provider to SMEs. From there we moved to an electronics prototyping open innovation platform. And then it just became an open innovation platform…so yeah (laughs). At one point, someone asked us if what we were doing could be used to evaluate data scientists. There was a lot of demand for big data researchers and data scientists, but not many recruiters were familiar with the field. We saw an opportunity there and realized that we could adapt our technology to do this, given the product and our background. So that’s how we entered the recruiting space.
When you started out with Ideatory, you were still doing your PhD while working part-time. Then what made you switch to full-time with impress.ai?
The work at Ideatory didn’t need me to commit full-time. We were doing well, but we wanted to do more to leverage our technical background. So we got together and identified this new, exciting problem that we could solve. And that’s how Impress was conceptualized. With Impress, we had so many ideas on what could be done. It was all very exciting. So I decided to quit academics and focus on this completely. It wasn’t that I didn’t enjoy my academic work, but this was something else entirely. I loved the freedom of having my own product.
And your journey with Impress only started in August last year?
That’s correct. August was when we got the idea. Then in November, we entered this Hong-Kong based startup accelerator called Zeroth.ai. By January, we were ready with a self-service product. The accelerator really helped…well accelerate things (laughs). They had industry experts who could tell us what would make our product saleable. We had to come up with updates every week. Demo day was a self-set deadline for us to achieve pre-decided targets regarding customer, product and funding. And we did manage to achieve all of those (smiles).
Could you explain to our readers what you do at Impress AI?
Basically what we do is help companies screen candidates better. Now, what happens is that once a company (the recruiter) signs up for an account, they gain access to an interview creation engine. It is a screening interview, where there are three levels of information. The first is a basic level, when you ask for resume uploads and stuff. At the second, slightly deeper level you choose what other information you want to ask – notice period, residency status or background information and so on. And the third level is for you to assess certain competencies such as teamwork, leadership skills, programming skills and so on.
And where does the chatbot come in?
So the interview creation engine is actually the chatbot creation page. After you select the information you want to ask in your job application, you just click Next and the chatbot is instantly created. You can embed this bot anywhere, on your own job page for instance. When a candidate clicks on “Apply Now”, the bot pops up.
So it’s not an app?
No, the good part is that the bot is platform-independent. It’s completely web-based. There’s no need to install or download anything.
Where does the artificial intelligence (AI) component come in?
The AI part comes mostly after the candidates submit their applications. We convert the transcripts to a set of scores on different competencies. A score on leadership skills, a score on programming skills etc. So what we are talking about is auto-annotation of data. As companies rate sections of the application, the machine-learning algorithm learns their preferences over time. In the long-term, it implies more and more automation of the screening process.
Apart from this, we also work to eliminate human bias from the process. Each applicant has a unique identification number. No kind of personal information, such as name, age, gender, religion etc is revealed. These are factors that unintentionally influence the hiring process in a traditional setup.
Vaisagh, from our dialogue so far it seems that you’re an academic who stumbled into entrepreneurship. How do you cope with the extremities and unpredictability?
Well there are highs and lows. And they can happen on the same day (chuckles). One moment you get a positive feedback from a new client that makes your day, and the next moment someone you’ve worked with for a long time decides not to continue. I think you just need to develop a thick skin and find your safe space. For me it helps that my family, especially my wife Aparna has been supportive throughout.
About the unpredictability, it’s not my thing. I like things to go as per plan. I am not a risk-taker by nature and never considered myself to be the entrepreneur type. So how I deal with the uncertainty is to make a list of things to do and then I feel better as I go on ticking each one off (laughs).
Has your PhD experience helped you?
Definitely. A PhD is like a marathon. You need to have the stamina to get through. At times, when you submit papers you get reviews that make you want to crawl into a corner. But at the same time you might get a review so encouraging and supportive that it puts you back on track. I think whether you’re doing a PhD or entrepreneurship, you just have to be extremely determined and self-assured.
How important do you think it is to find the right co-founder?
It’s very important. You don’t have to be best friends, but you need to know each other well. More than anything else, your value system needs to match.
Where do you envision Impress.AI in the coming years?
We are looking forward to grow the company, hiring employees and setting up a company culture in general. Apart from that, we intend to further our vision of eliminating bias from the hiring process. For that to happen, impress.ai needs to become a big movement that will come with large scale customer adoption, global scale and thought leadership.
And with that we come to an end. Thanks for a detailed and insightful interview that gave us a glimpse into the tech startup space, and AI at that. Before you go, what advice would you offer to wantrepreneurs (wannabe-entrepreneurs)?
Entrepreneurs are known to be highly passionate about their work. The motivations might be different, but they have this drive from within. Personally, I just tell everyone the negatives of doing it. Especially the uncertainty. We have had months where we couldn’t pay our salaries, but then we’ve had some pretty awesome times too. In a pros-vs-cons list, there are definitely more cons. But even after that if you feel the need to do it for whatever reasons, then most definitely go for it!
Dear Readers, I hope you enjoyed reading this. The following video will give you an additional demonstration on how Impress.AI works. Check it out!