Meru Cabs: Parking Bots and AI in the right spot?

Pratima H

MUMBAI, INDIA: The trunk space matters. It not only holds stuff that you can’t put on seats yet but also offers hospitality to tools, spare wheels and the miscellaneous. When some of that starts getting right under the hood, at an arm’s length to the core engine, and at a hairline’s to the heat of things; then one gets tempted to reverse the role of a pillion-rider. Asking instead of annoying takes over. Questions and not instructions start to flash headlights. In short, the guy in the driver’s seat gets to enjoy the accelerator once in a while.

Is that what Nilesh Sangoi, CTO at Meru Cabs could be feeling with technology? It has been a long drive after all since Meru started a journey to create a taxi segment that caters to the evolving transport needs of the Indian customer with GPS/GPRS-enabled cabs to redefine the space with a new ecosystem altogether.

Technology has been claimed to be an integral piston on this road enabling Meru (as the company states) to be the first taxi player to introduce the Carpool service in October 2015 for customers to share their personal cars, reduce traffic congestion and contribute to a greener environment. It also dovetailed with the RideShare service, which as Meru tells, witnessed tremendous success in both phases of the odd-even initiative in Delhi.

Since its launch in Mumbai back in April 2007, where-hence Meru has traversed 24 Indian cities already, it has also experimented with models of a versatile range – from the MeruGenie services in December 2013, to Meru Eve- a cab service exclusively for and by women in January 2015 etc.

Technology has been quite a lever in many other initiatives it seems – like a specialized service for differently-abled and elderly commuters that use disability equipped vehicles, a level of as much as 70 per cent of its booking happening from the mobile app, services like seamless booking and navigation for its customers.

Now the next frontier has beckoned it further – Meru is integrating its booking system with Facebook’s Messenger Bot.

This integration of their cab booking system with Facebook Messenger is a radical step to provide automated cab booking facility that will allow users to book cabs through key words, intelligent chat bot and conversational question- with answers popping through the service called Meru Messenger Bot.

Among other developments, Meru has recently announced a new round of funding of Rs.150 crore ($ 25 million) from Brand Capital with the intended focus on providing quality service based on a sustainable business model. In the organised taxi-hailing market of India that tots up somewhere around Rs. 7,900 and can rev up to Rs. 1 lac crore by year 2022, Meru faces a roadmap full of rainbows and cliffs.

What would make technology help it ride on the right side and what more lies under the trunk? Now is the time when many questions are peeking from there and other places – specially with the noise that autonomous vehicles (and their crashes), the other side of bots and impacts of uberisation are making. Let’s get in the pillion seat and get a better view of the road as Nilesh S steers through them.

A bot to interact with customers! What triggered this idea? Specially when everyone else is trying to catch the app-first wagon?

We have been looking at the success of WeChat in China. It’s amazing how something like that can be an end-to-end commerce platform for users. Globally, no one has replicated that model yet. While the app-only or app-first model is getting hot, the bot opportunity is actually huge. So far there are limits to what US Chat-platforms like WhatsApp can achieve from a business context.

We tried that route also but in the last few weeks, opening up of APIs and Facebook’s announcement unbolted new doors. There has been, in parallel, a lot of improvement in cognitive computing which is helping the bot opportunity. Like colloquial translations have got better and so has the ability to interpret text. We were fairly ready to tap chat platforms so we were quick to absorb this new option that came in with Facebook. In a matter of ten to 12 days we got rolling.

Bots have some advantages over apps: Nilesh S

And what it means essentially is?

Meru Cabs has become the first cab services provider in the world to go live with implementing Facebook’s Bot services. What we plan to do is to widen our customer base by providing convenient and seamless cab booking access to its users. Now users can simply search 'Meru Cabs' on Facebook Messenger under ‘Bots and Business’ section and begin interaction. After the initial mobile number verification (one time), the user would initiate a normal chat with Meru MessengerBot to book a cab and once the location is shared on the chat box, the user is then directed to available cabs for the trip with the ETA (Estimated Time of Arrival). Also, Meru Messenger Bot responds on real time basis with the cab details along with the chauffeur’s mobile number while enabling the user to locate his/her cab and check the fare details.

How has the ground felt so far?

Initial use cases have been around booking and tracking of cabs, cancellations or fare checking etc. Over 400 people have checked it and we have seen good traction. But it’s still early days and bots won’t replace apps overnight. Though nothing can be said precisely as of now, bots seem to have the potential to get big in the future.

Would it be a case of bots vs. apps or bots vs. humans? Do cost gains make up for replacing humans?

Talking of humans, it would still take some time to replicate the feeling and human response quality. A lot of research is happening and even AI scientists seem to feel that a human-level feel is 13 to 14 years away. I would say that the equation is more about experience than savings. Savings will come over time but experience cannot be frivolously dealt with.

As to apps, the experience differs on the aspect of structure. In an app, the creator decides the interaction and flow but a bot gives you freedom on what you want to say and when. It gives context and appropriate response based on that. Despite initial euphoria, people cannot handle too many apps in the practical sense. Many installed apps never get used. Discoverability of apps or promotions is also an issue. Bots can be used in several dimensions.

Would examples like Tay amplify the risk side of bots?

Anything which is not supervised is easy to corrupt. With these incidents the learning points and scale of learning can be huge. Taxi use cases would be different from the other contexts we are talking about here. No one is going to ask a bot about weather. Meanwhile, am sure that researchers will find a way to handle unsupervised learning. In our case, there’s an element that integrates at the back-end depending on NLP (Natural Language Processing) churn telling what the customer wants. Customers and APIs will become important factors and we will see how it works.

You have been recognized by ‘The Limca Book of Records’ as the largest fleet taxi operator in the country. With over 40 million trips across India or titles like India’s Most Favourite Cab Service (by TripAdvisor Travellers’ Choice); how much have you seen the industry change? Is it too soon to get ready for AVs (Autonomous Vehicles)? Or are we jumping the gun with driverless-taxis?

This segment will shape as per use-cases and research innovation. In India, the challenge would be different – infrastructure, road indicators, tracks etc specially on intra-city roads. On their own, AVs can drive well but when humans intersect then questions and doubts can pop up. In fact, it has been predicted that AVs can actually reduce the incidence of accidents by a huge factor. Like a McKinsey report informed well: The very annual cost of roadway crashes to the US economy was about $212 billion in 2012 and if one takes that year as a reference, advanced ADAS and AVs can be seen reducing accidents by up to 90 percent with potential savings of some $190 billion.

The level 4 of AV is actually the utopia and the safest one where they are completely unguided without any human intervention. Right now, most offerings have a level 3 version. Like cruise control, lane guidance, ABS etc. Auto companies are taking an incremental approach from level 3 to level 4. But by 2019, as some experts have predicted, there will be visible progress. Driving would become so redundant and easy that sitting in a car could actually mean sitting back and having coffee. But this will happen incrementally and a lot would depend on the learnings we have and the approach or speed regulators take.

Is it a fuzzy time that way? On one hand there are mishaps like the recent Tesla crash, or the earlier Google one; and on the other hand auto majors like GM are tying up with players like Lyft?

Would AVs take over our roads soon? Not too likely. But the auto industry has not changed too much in the last 125 years except for the drive to build more powerful vehicles and new models. Now the last ten to 15 years give a glimpse of a new approach trickling in. Starting with the DARPA contest, then to university collaborations, then to Google etc joining the fray; today every auto company worth its salt in looking at these new technologies seriously. Safety-wise AVs have performed well in normal conditions, now would the same apply in extreme conditions?

Specially as a lot depends on data that comes from AI, Machine Learning, sensors etc. The regulatory part would have to shape well on many fronts – like accountability of accident or insurance etc. Those parts will still take four to five years. If you recall McKinsey’s prognosis, through 2040, on-highway trucks could more likely be the first vehicles to feature the full technology on public roads. Imagine the questions of jobs and industry issues that will surface then.

Talking of future, what else is expected at Meru in terms of technology coming out of the bootspace?

Algorithms would be a big and major thing here. We are doing a lot of work there and also exploring business-oriented AI cases. We have built good systems and will progress fast.