Deciding to become an entrepreneur is ridden with risks. People give up their well-paying jobs and plunge deep into their business, once they know what they want to do. But for Ashutosh Saitwal who had 25 years of work experience mostly in the IT product space, the situation was a little different.

Says he, “When I founded our startup, I knew what I did not want – I did not want to be in the IT service industry. I had a lot of experience in the IT product space as well as the financial industry. So, I thought I could build a fintech product. That’s it.” Not really enough to get headlong into a business, but for Ashutosh it was good enough. And so, the journey of KlearStack began in 2018.

Ashutosh of course was willing to put his money where his thoughts were and began by investing entirely from his savings. Though he won’t reveal how much, it “was in the region of several crore rupees.” Given that product development is a capital heavy idea, Ashutosh thought that providing some kind of service would help hedge part of the investment. And so, he started offering Robotic Process Automation and Machine Learning services to various clients. “It’s a little faster to get clients for services rather than for a product.”

Providing services turned out to be fortuitous for Ashutosh’s product journey. “since we provide automation services, seven out of 10 clients asked us if we could help them with the data their invoices held.” And that turned out to be the bingo moment for Ashutosh. “If seventy percent of my clients need some help with their invoices, there must be something to it.”

Of course, 25 years of working experience had brought maturity and Ashutosh didn’t jump straight into developing an intelligent document processing (IDP) product for the need that he saw. “I read up a few analyst reports, mostly Forrester, Everest, and discovered that in 2018 alone the USA generated more than 500 billion invoices and receipts.”

But India? “We had bid for a government project in those days, which we did not win, but the experience helped us gauge the number of documents that could come up in the domestic market. The government wanted a system that could extract data from photos of bills, read that data, and give it to the GST portal. The aim was to plug fraudulent tax avoiders. Talking to the government authorities and some experts, we gleaned some information. The government was looking at tens of Crores of invoices per month!

And though kismet was being benevolent, there was some tough work to do. “I had a small R&D team and given my work experience, we decided that rather than build a generic platform, we would focus on two or three use cases within the fintech IDP and become good at it. And then look for more opportunities to become a more intelligent document processing platform.” And so KlearStack began its work.

The first use case was consumer loans. If a consumer wants to buy, say a TV costing Rs. 2.5 lakhs on EMI, there are a lot of documents that have to be processed by the lender. At loan eligibility determination stage, there are KYC documents and the loan application form.

Instead of manual data entry from these documents, the bank salesman simply takes photos of the KYC and the application form on their mobile app. These pics of the documents are uploaded onto KlearStack cloud and KlearStack interprets and extracts relevant data using our proprietary computer vision, NLP models trained using Machine Learning. This interpreted data then is pushed to the bank’s loan processing software. The loan is either approved or rejected and the status then goes back to the bank salesman’s mobile app and in 10 mins the job is done. This otherwise would take a few hours.”

The next step in the loan transaction is the consumer filling and signing an auto-debit mandate form. Here too, KlearStack built a new module that helps a finance company that needs to process such NACH mandate forms.

The next step in the consumer loans lifecycle is where the bank or financial institution needs to process the insurance policy document and extract data from the same. “If a customer buys a car or a two-wheeler on EMI, then the bank has to ensure that the vehicle is insured. This involves documentation. The consumer shares the insurance papers with the salesman. Again, bank needs to extract and validate the details like engine number, chassis number, insurance value, etc. Here too, the bank salesman simply has to take a pic of the insurance policy document and upload it to KlearStack. This is then matched with the loan data and put on the onward system of the bank. Now the bank is satisfied that there is no fraud in the deal.”

The second use case for which KlearStack developed their product is accounts payable invoices. “Most large companies have thousands of invoices generated by their vendors. Generally, they do not follow a fixed format of invoices. This makes getting information out of these invoices tedious. Data has to be entered manually into the ERP. Companies had help, but that was a template-based system, which is very time-consuming and effort intensive.”

KlearStack built a proprietary product that uses Optical Character Recognition (here they use a 3rd party engine in addition to KlearStack OCR) with proprietary Computer Vision, NLP, and the deep learning models. Says Ashutosh, “we can even monitor our clients’ mailboxes, pull out invoices from email attachments or email body, process them to extract data like invoice date, line items across multiple pages, etc. This well formatted, structured data is then pushed into the client’s ERP.”

To market, to market, to sell a cool thing

“Our product development was going on simultaneously and in Sept 2020 we were ready to hit the market.” Ashutosh did not approach those same clients from whom he got the insights into his product development. “We used our website and optimized for search engine. We consciously focused on rich content that helped us improve our online visibility and through that got customers.” IDFC First Bank was their first big client. And soon more followed. In fact, this SEO marketing helped bring in multiple investor inquiries too!

When it rains, it pours for KlearStack

Recently, KlearStack achieved monthly breakeven. That is when Ashutosh felt that the company has achieved a product market fit and decided the time was right to raise money to fund further growth. Says Ashutosh, “Investors saw that our product had a good market. Our content marketing did all the talking, and four investors approached us. But we chose to go with two of them because they got in smart money.”

This was Virenxia group from Dubai and ITCube Solutions. “The other two investors came in with cash, but these two could offer us strategic value. Virenxia wants to license our technology for use in their agri-tech and sustainable rural development platform, and ITCube has started promoting our product in the US.” Ashutosh has given them a “minority stake” after some hard negotiations. “Negotiation is more of an art than science,” says he. “Though they knew the market opportunities, our unique solution, investors always want more. But thankfully for us our competitors raised multiple rounds of investments at astronomical valuations, so we used this data to an extent to support our hypothesis.”

Ashutosh is using the money to expand his services to the US, expand his team, and build more depth in products. “When I started out, I took a calculated risk, but I was sure that we would do something meaningful. We are not there yet, but this sure is a step in a long journey.”