In August 2018, I had written a story about how Bikram and I bootstrapped a profitable AI consultancy. In startup timeline, a couple of months is equal to a couple of years. And the last few months have been nothing short of a roller coaster for us. From not getting paid by a client to closing our tech consulting business, to getting rejected by YC after being shortlisted, to building our first SaaS product, we have struggled and created our own path.
If creating a sustainable service company was milestone 1, then we are currently on milestone 2 — launching a vertical AI SaaS product. And the goal with this article is to help future founders who are already in the trenches make the hard choices.
Running a technology consulting business is tough. What we realised is that it is not scalable either. Converting a qualified lead to a project takes about 3 to 4 months. Client requirements are never the same & there is a very little scope of standardisation. Hence, every time you end up talking to multiple stakeholders, understanding requirements, drafting multiple proposals & negotiating deals. Add to that the probability of client dropping out at any point of the process and you get a very leaky sales funnel with a lot of wasted time.
Once the project starts, you realize that the scope is much larger than you initially budgeted for, thus eating away at your margins & time. In one of our projects, we had the unfortunate experience of a client refusing to pay us for the last deliverable due to mismatched expectations. What will you do if your client refuses to pay you? Well, you can't do much if you are a startup. Just learn to create much smaller milestones next time onwards & get paid weekly/bi-weekly.
From those 8 months of delivering AI solutions, we realized that we would like to solve one problem and solve it really really well. So we decided to switch gears!
The amazing part about starting up is that there are no playbooks. Each journey is unique in its own way. If you are always on the lookout for good opportunities, you will find your way. For us, that was in the form of YC Startup School where we had the privilege of being mentored by Neil Joglekar (YC alumnus, Stanford & Google). The advice Neil gave us during our first office hour was “Your KPI should be the number of weeks to not being a consulting business” and it struck a chord with us.
We made a list of problems that we have faced in our previous jobs and in life outside of work. We ranked this list by size of the market, our capability of executing and most importantly, our passion for solving the problem. In the end, we decided to focus on the problem of digitizing documents.
After conducting multiple user interviews in India and USA, we realized that this is a huge problem. Imagine all the documents that a company receives — invoices, receipts, purchase orders, delivery notes, forms & contracts. All of these need to be digitized for the business to function. So far, most of this work is done manually or using unintelligent OCR. And with the advancement in deep learning, this space is about to be disrupted massively. Bookkeeping, in particular, is one such use case that every company needs to implement though it only adds to the cost & is not scalable. That was the inception of Docsumo.
Over the last 3 months, we stopped taking service projects. We communicated to the team about the shift in focus and started devoting all our time to creating Docsumo, a vertical AI SaaS web app that automates bookkeeping and data entry.
Startups don’t die because of bad ideas or founders, they die because they run out of cash. Cash is the lifeblood that keeps the organization pumping. And building a product is one of those exercises which sucks cash for months before generating revenue. We knew this and hence we followed in the footsteps of wiser founders such as Jason Fried (Basecamp) and Ben Chestnut (Mailchimp) who ran consulting companies to fund their products.
As we had initially hoped, the technology consulting business gave us enough cash to build out our MVP and test our product with users. Our cash reserve gives us the advantage of iterating a couple of times & get closer to product market fit. This has helped us to validate our hypothesis before taking on investor funding. Yes, we might raise funding in the future but that would be mainly to scale and grow the product.
Imagine being able to convert documents to digital data with the click of a button. That is the future that we are trying to build and we are excited to bring this product to market. I will be creating a demo and writing more on the technology in the next article. So stay tuned & good luck!
In today’s dynamic business world, filing and archiving official documents in the digital form makes it handy, and works wonders in the future or in unforeseen circumstances.
Optical Character Recognition (OCR) is the technology to convert an image of text into machine-readable text. It is the underlying technology for various data extraction solutions including Intelligent Document Processing. However, OCR is not smart enough to figure out the context in a document - it works simply by distinguishing text pixels from the background and finding a pattern. This limitation could cause inaccuracy in captured data that could directly impact the output of your data extraction model.
Accounts payable is a key financial function for any business. Corporations can have thousands of suppliers; even for relatively smaller businesses, the number of suppliers could be in hundreds. All the invoices they receive from these suppliers come in multiple formats, layouts, and templates - some semi-structured, some unstructured. Therefore, firms expend time and resources to capture invoice information through manual data entry and verification of accounts payable. Manual data entry is not feasible in the long run, definitely not on a large scale. Before we talk about how intelligent invoicing solves the problems associated with manual invoicing, let’s discuss the challenges in much detail.
As most of an organization's information is available in an unstructured format, processing it requires an automated system that can handle documents with minimum human interaction. OCR is one such technology, but its scope is limited as it requires human interaction and is highly dependent on the layout and structure of the document to be processed.These limitations are overcome by Intelligent Data Extraction.Using artificial intelligence, the Intelligent Data Extraction technology extracts data from documents and transforms it into useful information through the extraction process. It functions as a singular tool for extracting information from any type of document and aids in optimizing company operations.