This website requires JavaScript.
EditorPicks

"Almost impossible to fool someone in today's world"

Engineer.ai's co-founder makes a case for the startup in the light of The Wall Street Journal and The Ken articles accusing the company of sharp practices.

Sep 2, 2019 by Moulishree Srivastava
"Almost impossible to fool someone in today's world"

Engineer.ai isn’t having a great year. August was particularly troublesome. Founder Sachin Dev Duggal has been fighting a lawsuit filed by a former employee over wrongful termination.

Earlier this month, the Wall Street Journal (WSJ) published an industry story about AI startups hyping tech to increase valuation and get investors’ attention, only naming Engineer.ai. The company believes the former employee reached out to a WSJ editor who he was acquainted with and orchestrated the article. A couple of weeks later, The Ken did another piece on how Engineer.ai is bluffing about not just AI capabilities, but balance sheet, clientele, partners and employees.

The US-based company began operations in 2013 selling cloud services from multiple providers. Three years later Engineer.ai started building a product to create applications, which has now metamorphosed into a platform to ideate, build, run and scale applications. With offices in Los Angeles, London and India, the 195-people strong company says it builds software applications for its clients using a combinatorial network of 62,800 contract developers, automation and AI.

In a candid interview with The Passage, Duggal spoke about their side of the story. “We are not selling AI services, we are selling a platform that is powered by human-assisted AI. We are not one of those companies which are using AI for the sake of it--we only use it where we absolutely need it,” he said.

Edited excerpts

The Passage: Engineer.ai has been in the news for all the wrong reasons. We are interested in the things you did right to reach where you are.

Sachin Dev Duggal: We started our journey with the cloud offering, CloudOps, because we wanted to make sure that we had a recurring part of the revenue up and running. Building software is transactional revenue. If you are going to build things for customers and they are going to leave and host or run or have someone else manage their environment, that is not very helpful, because then you will have to re-acquire those customers. When you are bootstrapped, you can only do small things at a time. So we focused on the top of the funnel problems first. In 2016, we did an early version of what we call Builder studio to figure out what is it that customers are after. In 2017, we realised the first version, which was based on people wanting themes, wasn't strong enough. So we switched to offering custom solutions--to build a software project. We launched an early version in 2017 and did the first public release in November 2018. We had bootstrapped the company from the start and reached an aggregate revenue of USD 60 million before we raised USD 29.5 million in Series A last year.

The Passage: Where exactly does AI come into play on your platform?

Sachin Dev Duggal: First two areas are pricing and timeline. We also use facial recognition to ensure the developer who takes the qualifying test is the same one doing the actual work.

Another example is Builder cloud, where we do usage and churn prediction. We started with machine learning model for usage prediction and we were using it for a while, but we were getting only 70-80% accuracy. Then we sought the help of a traditional statistical model and that actually gave us accuracy in high 90s. So it turned out to be better than a machine learning model.

We currently have six, seven people who understand intelligent systems. We plan to raise the number 20-25 before December. The thing is, we had to get to this point before it made sense to do these investments. We spent the last couple of years collecting a vast amount of data, and now we have to use it for our upcoming projects.

And in most things, the general progression is you do it manually, then you do programmatically, then automatically and then intelligently. You can never skip the process. If you do, you build the wrong solution.

The Passage: The WSJ article on Engineer.ai was based on the words of a former employee, who has also sued you for wrongful termination. Your thoughts?

Sachin Dev Duggal: We have active litigation with Robert Holdheim, so we can't comment much. But we vigorously defend our position. The WSJ article said we had told our investors that 80% of our platform is automated when in fact we had made it clear to our investors that we are at 40% of the journey. However, WSJ chose to ignore our statements.

Clearly, we do not have everything right. We are fixing a bunch of things. Sometimes left hand doesn't know what right hand is doing. We hired 70-odd employees in the last six months. We have a really wide age group of people older than a specific former employee. We take pride in our cultural diversity - 50% of the company is Indian, 30% is British, and 20% is American.

The Passage: Do you think some tech companies talk up AI capabilities to attract investors?

Sachin Dev Duggal: AI startups broadly fall under three stages. There are folks who are doing research, folks that are doing AI as a service, and folks offering services powered by AI--companies like us.

Broadly speaking, when you ignore research and look at AI-as-a-services company - what you are really talking about is a new level of prediction system. The statistical models we are using or the open-source solutions that are there, they are going to make their way into every company. I think companies that are saying AI and don't have any data scientist and are not even using any outside system, you have to be a little wary about them.

There is another part to it. We are a lot harsher on tech than we are on other industries. Does Red Bull grow you wings? Just because you have Nike shoes doesn't mean you can run 100 m in eight seconds. All these companies, they never have a conversation about how they market themselves. But for tech, we take the exact extreme opposite view, that you should basically say what is there right now and forget the vision.

We are selling a vision, we are selling a feeling and emotion. Tech companies get chastised because everybody assumes the investors are silly and they are just going to invest behind a company because it says it has AI.

And there isn't actually a general consensus on what is AI. All automation starts from rules. When you get enough usage and enough data, then you start deriving the rules from the data. And that is where you switch to a deep learning model.

I know how many frogs we had to kiss before we got money. Investors are not unsavvy. It is almost impossible to fool someone in today's world.

https://cdn.thepassage.cc/filters:quality(70)/public/image/2019/09/02/Guest_Article.jpg

The Passage: Do you think investors are able to separate signal from the noise when it comes to AI claims?

Sachin Dev Duggal: My question is different. Who is a qualified investor?

If you have got a bunch of investors who are not savvy, and if you have got very broad, nuanced terminology like AI, it is a recipe for disaster. Ten years back you did not have investors like you have right now. So all the money was really smart.

If you are investing in a company on the seed stage or a series A stage, it is without conclusion they have not built out their entire stack. Hence it is called Series A.

When you are investing in seed-stage, you are investing in the vision. What investors are evaluating at the seed stage is the market problem and if yours is a viable product. When they are evaluating a Series A company, say a research company, doing deep tech---then they are evaluating how efficiently you are solving a problem and whether you have penetration of the problem.

So in none of these few cases, there is a risk of confusing people, because, for most people at seed or A, it is all about vision or product-market fit.

The Passage: How did the WSJ article impact the company?

Sachin Dev Duggal: I think what it has taught us is to have a far more robust cultural interview process. We went from 25-30 to 200 people really fast. When you hire quick, onboarding and communication are really important.

The WSJ article was meant to bring us to the knees. And it didn't. We responded; we dealt with it, we replied and we gave our point of view.

So we were asked were we trying to make us sound sexier. But I ask you, which CEO tries to make his company sound unsexy. If you are not creating a brand strategy, you are not marketing. Also when you do an industry story you need to mention more than one companies, and we were the only company mentioned in the article. We are now asked to behave like Apple and Google, but we are still a product. So we don’t have the processes in place to deal with it, we don't know what data to share, what not to share, we don't know what to say, what not to say---the only thing we do know is we can't explain the technology on a phone call. Everything else, we have been learning by the minute. I am also shocked at the number of people who have written about us without speaking to us.

We haven't got any emails from them, and yet they say they have tried reaching out to us.

On the flip side, everybody knows us now.

https://cdn.thepassage.cc/filters:quality(70)/public/image/2019/09/02/Guest_Article_1.jpg

The Passage: How do you plan to move forward?

Sachin Dev Duggal: We knew we had to do a better job of communicating two months ago. From the strategy and product perspective, nothing has changed.

We are doing some big releases in November. One of them is about allowing people to experience their ideas within minutes. It is about how we can build the next layer of automation to accelerate the development cycle by at least 35%.

The Passage: The Ken article said the writer got an unrealistic timeline and base price (9 years and USD 251,700) when he tried to build a replica of Flipkart app on Builder.ai. How do you explain the anomaly?

Sachin Dev Duggal: The writer of the article chose one of our templates and added 383 features, while the standard template for that app had only 30 features, was priced just under USD 25k and delivered in no more than 38 weeks.

It’s not a realistic scenario. The intent seems to be to create a perception that the system doesn’t work. Especially when you see the features like “Paid Content” which is probably not on Flipkart's core offering! The pricing of the features is done by AI.

The Passage: The Ken article points out the reported revenue for FY18 from the Builder.ai platform is just USD 350,000 and 97% of its revenue comes from cloud services. Your comments?

Sachin Dev Duggal: We only ever talk about total revenue and report this in our consolidated revenue from global operations. We gave Ken details of five entities, but Ken used the revenue from one entity; especially as it is not a parent entity. Any conclusion drawn from the analysis of filing of one Indian company will definitely not lead to a correct inference of global platform revenue. We are a global platform sold through different entities globally.

Additionally, we also provided the writer with a letter from a CA firm who did our financial due diligence last year. This was also ignored by the writer. The letter confirmed we had generated USD 23.08 million in revenue for March 2018.

We made it clear to the writer that we are not reselling AWS. We are in fact a multi-cloud aggregation & DevOps platform. Unlike a reseller we don't simply buy and sell on a 1:1 basis; and the take rate is a notch higher because of our approach to aggregation and value-added services we provide to clients.

The percentages referenced are only for a single Indian Entity and are not reflective of the group. The Builder revenue is in the millions, and certainly in double-digit percentage figures, but we do not separate out platform revenues.

The Passage: The article has raised a question mark on your USD 100 million-exit from Nivio and chalked up Videocon’s investment in Engineer.ai to nepotism. Comments?

Sachin Dev Duggal: Nivio raised money from many partners including Videocon.

Many private companies do exits of shareholders without any public record. In many cases, you sign strict confidentiality agreements not to discuss the details. There was an exit for many shareholders, and the company’s valuation was USD 100 million. We can furnish financial transaction details off the record as they are private in nature.

The writer of the article had been in touch with me many years ago to collaborate with Nivio. This was left out of the story.

The Passage: Do you plan to take action against WSJ or The Ken?

Sachin Dev Duggal: We are building a company and we intend to focus our energy and resources to keep building and growing. Our best response would be our continued success.

Moulishree Srivastava

Moulishree Srivastava is a Bangalore-based tech journalist. She focuses on emerging Indian startups and unicorns. She can be reached at moulishree@thepassage.cc.

Follow Moulishree Srivastava