The slow, light touch of AI in Indian Saas

Looking at him from behind, Sridhar Vembu, in the picture, has a halo around him. It’s the lighting in the room. But for several entrepreneurs, he’s a demigod. Not for nothing. Building a profitable software as a service company that makes close to half a billion dollars in revenue, without raising venture capital, is a superhuman feat. Vembu has done that at Zoho, the company he founded more than 20 years ago.

We’re at a beach resort in Mahabalipuram, a little off Chennai — India’s software as a service (Saas) capital. Over 100 founders are in attendance at the fifth edition of SaaSx, an annual conference where entrepreneurs like Vembu share their lessons. Some have come on an overnight bus hired by the organisers, some others have flown in.
I was on the bus, named the ‘Sassy Bus’ (a pun on SaaSx), which started a little after midnight from Bengaluru. Early the next morning, we reached Mahabalipuram. At the pit stop, while we waited our turn to use the bathroom, Pravin Singh, a volunteer at iSpirt, the software products thinktank that is organising SaaSx, has struck up a conversation with entrepreneurs around him.
“What’s your ARR,” he asked, pointing at them one by one.
For the uninitiated, ARR is short for annual run rate. It’s the money you’re on track to make at the end of the year. No one in the group of six-seven founders had a run rate of over $1 million. “That’s why we’re in the bus. The ones who are making over a million will fly in.” We broke into a laugh.
It sadly true, though. Most Indian Saas companies haven’t crossed the $1 million mark after several years in business. And, even as they struggle to perfect the art of selling to the world, raise money, or strike partnerships, there are dark clouds before them: a new breed of companies that use artificial intelligence are threatening to upend their plans.
These new companies, mostly emerging from markets like the US and Europe, are not only automating tasks but also have the potential to sell cheaper – wiping out the cost arbitrage that Indian Saas companies have gained from. The classic Saas business model– which charges customers per licence– is also moving to an outcome-based pricing model.

“By no means has Saas caught up in India from an AI perspective. I’d argue that we haven’t even reached the original promise of Saas in India,” says Ashwini Asokan, the co-founder of Mad Street Den, a Chennai-based AI-first company. Asokan is right. Barring a few like Zoho or Freshworks, Indian Saas companies are yet to make it big in the global market.
India has roughly about 500 Saas companies and a majority of them have an average run rate of less than $5 million. Back in 2016, a study by Google and Accel Partners had projected that Indian Saas businesses can potentially serve 8% of the global demand by 2025, making it a $10 billion industry. By 2020, the study projected, Indian Saas will turn in $2.5 billion. (Accel Partners is an investor in Sourcecode Media, the company that owns FactorDaily.)
But, the market is fragmented. Another report, the India SaaS Survey 2017, by iSpirt and investment bank DCS Advisory, has a telling data point: the median ARR, among 59 survey participants, was just about $750,000.

Hurricane AI

Computing power, data and newer methods of training machines have revived the field of AI. This time around, it’s more enterprise ready. Gartner predicts that the business value derived from AI will touch $1.2 trillion in 2018. A large chunk of this, about 46%, will come from what are called virtual agents – software robots written for specific tasks.
Virtual agents won’t need customer relationship management software built for humans to and understand customer data. They will interface directly with databases and customers.

It is already happening at business process outsourcing (BPO) companies, as FactorDaily reported in this March story. The example of EXL Service using a virtual agent to summarise doctor diagnosis and prescriptions in an insurance claim process stands out: the AI-driven software takes less than a minute to do the task that would take a human agent two to four hours. Aren’t Saas companies on the same trajectory is the question.
“Indian SaaS companies that are building products for workflow productivity can be at risk,” says Chintan Mehta, partner & fellow in residence at iSpirt. Companies that build products that have artificial intelligence baked in will have a better chance of selling as compared to those who don’t, he adds.

Agara Labs, a startup which we wrote about earlier, is a good example to illustrate this point. The startup is building an AI-based solution to address customer queries and is already working with one of the world’s largest consumer products company. Another example: Beautiful.ai is a product that helps users create presentations. Many tasks that a pixel pusher would earlier do to make a presentation look good are now relegated to AI. The product was featured five months ago on ProductHunt and is getting rave reviews.
“If you are a software that is serving people for a function are at risk. Find a way to serve that function instead,” says Mehta.
Indian Saas companies have for many years gained from lower developer costs. According to one estimate, Indian developers are at least four times cheaper compared to their US counterparts. The Google-Accel study of 2016 identified many areas including healthcare, logistics, sales/ CRM, productivity and collaboration, services, support and helpdesk, data management and visualisation, where the demand is primarily met by US companies, are ripe for Indian companies to target.

Lower costs meant you could easily clone or make better products and sell them abroad cheaper than a comparable product. That equation is set to change with AI coming into the picture. AI talent is scarce and expensive. Companies like Google, Facebook and Amazon have a natural advantage when it comes availability of data and talent. This will also push up the cost of developing AI-first products in countries like India unless it invests in training high-quality talent.
Also see: India’s AI talent is bottom heavy

More open to outcomes-based pricing

To add to the growing list of worries: a change in the business model of selling software is imminent. AI-first companies can charge customers based on business outcomes whereas traditionally, the Saas business has worked on the ‘per seat’ pay-as-you-go model. Enterprises like outcome-based pricing a lot more than buying recurring licences. “Revenue models that are based on the number of operational users seem to be threatened. Revenue models based on end-users or end-user transactions seem safe for now,” says Sangeeta Banerjee, CEO of Apartment Adda, an accounting and management software company for apartment complexes.
Even successful companies aren’t immune to the threat posed by AI-based companies. Global leaders like Salesforce are doubling down on AI, baking the technology into different solutions the company sells to its 150,000-odd customers.

It’s an early trend but Indian companies such as Zoho and Freshworks have also started investing in AI. Zoho has launched Zia, a conversational bot that can do sentiment analysis or learn new skills. Besides hiring AI ready talent, Freshworks has acquired companies like Chatimity, Frilp and Joe Hukum to build its AI stack. Its CRM solution Freshsales, now ships with built-in AI features.
“They will clearly have a lead,” says Mehta. But the new wave of AI-first companies have the potential to grow faster compared to the first wave of companies, he adds. “Can’t really say whether they will overtake. But either outcome is desirable.”
Others point out it is not judgement day yet. Suresh Sambandam, the founder of workflow management software company Kissflow, has a grounded view of AI. “AI is good and we shouldn’t ignore it. But it’s not like we should leave everything that we are doing and jump ship,” he says. Kissflow uses AI inside its workflow management solution to optimise a newsfeed of sorts for its users. But for him, it’s too early for the industry as a whole to focus on the AI narrative.

Sambandam’s hypothesis is simple: Silicon Valley focuses on the “sexy stuff” like AI and autonomous driving. Israel is focused on cyber-physical systems, embedded chips and such. India has built a $150 billion information technology services industry on the back of domain knowledge and applications that were earlier custom developed can now be productised for the cloud. “You combine that with the fact that Silicon Valley is not working on SaaS because it’s not sexy, India becomes the third ecosystem which has the biggest opportunity,” he says.
If one goes by Sambandam’s experience of 20 years as an entrepreneur, Indian B2B Saas companies may yet find success in the niches that are often ignored by Silicon Valley entrepreneurs even as AI looms on the horizon.