(From left) Derick Jose, Krishnan Raman and Srikanth Muralidhara are trying to flip the usual model where data scientists solve engineering problems. (Photo: Jithendra/Mint)
(From left) Derick Jose, Krishnan Raman and Srikanth Muralidhara are trying to flip the usual model where data scientists solve engineering problems. (Photo: Jithendra/Mint)

An AI startup keeps mega-machines at peak performance

  • Bengaluru-based Flutura’s AI product helps global giants like Shell and Henkel move into the age of Industry 4.0
  • Flutura is also building a separate product to enable engineers to apply AI without being experts in data science

BENGALURU : Inefficiencies can creep into oil drilling rigs in multiple ways. Changes in the soil structure underground, for example, may require adjustments in the revolutions-per-minute, torque and pressure on the drill bit. An intelligent machine can do this round-the-clock better than a human.

Artificial intelligence (AI) could help with smarter decision-making. Trying to bore through a rock that’s too hard can damage the drill bit, which would entail a shutdown to replace it. But going around the rock may be inefficient if you can drill through it by applying more pressure. Such contretemps can add to the expenses.

“Every additional day beyond the planned time for drilling an oil well costs $75,000 to $5,00,000," says Derick Jose, co-founder of Flutura, whose AI product Cerebra helps oil and gas companies increase the uptime of their mega machines. Excess capacity is another money guzzler. Oil companies maintain standby fleets of critical equipment to avoid stoppages that drain money.

Flutura helps these companies reduce replacement or repair costs as well as keep a lower buffer stock of these expensive mega machines, explains co-founder and CEO, Krishnan Raman. Cerebra does this by analysing sensor data like vibration and heat from the machines, which helps maintenance crews fix problems before they spiral into a breakdown.

WORKING ACROSS VALUE CHAIN

Flutura’s first foray into the oil and gas industry in Houston, back in 2013, was a safety play. After all, a blowout can have disastrous consequences for an oil company. BP put the tab of its Macondo underwater oil rig explosion in the Gulf of Mexico in 2010 at over $62 billion.

With so much at stake, an AI predictive tool based on machine learning from sensor data to take corrective measures in time seemed like the best use case for Cerebra. But Flutura soon found that managers were more preoccupied with everyday matters like increasing output and revenue while reducing delay and loss. “The core business metric of an oil company is feet per day," says Raman, referring to how much a drill advances.

So the startup adapted to use cases whose immediate economic value to the customer was easier to see, such as preventive maintenance and operational efficiency. From there, it has reached a stage where Flutura has a foothold across the oil and gas value chain with multiple business use cases. It has also moved into the adjacent industry of specialty chemicals.

Henkel, for example, uses Cerebra in its Shanghai Dragon Plant, the world’s largest adhesives factory, manufacturing adhesives for things like sticking the wings of planes. That means the minutest of flaws detected after a batch of adhesives is processed consigns it to the trash dump. The wastage is huge in manufacturing specialty chemicals, and it’s one of the biggest costs.

But what if an AI system—modelled to read data from the chemicals at different stages of the process—can predict the quality of the final output? The batch can be corrected or junked midway, saving time and money.

This is what Cerebra enables Henkel to do, and today it is deployed in 60 of the German company’s manufacturing lines in 10 countries. Flutura has also moved into the catalyst manufacturing industry, which uses a continuous manufacturing process, where periodic sampling and prediction become more critical.

The third complementary industry Flutura has entered is heavy equipment manufacture, where it has customers like Hitachi. Jose sums up the strategy: “At the macro level, our mantra has been two-fold. One is to deeply verticalize and second is to globalize."

It took a few pivots to get strategy right. At the outset, power utilities seemed ripe for disruption with an AI product. Flutura tried everything from smart meters to energy retailing before finally deciding to move on.

“We found a mental model issue," says Jose. “On the power generation side, inefficiency was accepted, so there was no point showing all these bells and whistles. And on the retail side, there were no margins."

It was after such trials and errors that Flutura zeroed in on three verticals: oil and gas, process chemicals, and heavy equipment—all capital intensive sectors, where companies have to relentlessly drive down costs to survive. “We saw these as being ripe for adoption of our product," says Jose.

FOCUS ON THREE VERTICALS

For Flutura, focusing on three verticals and drilling deep into them is paying dividends. It enables them to discover multiple use cases with demonstrable economic value in each of the verticals.

“Domain understanding was as important as our underlying technology," says Raman. A $7.5 million series A funding round in 2017 helped. A VP-level hire came in from the oil and gas industry to focus on solutions. “Earlier we had generic data scientists. Now we also have people from companies like GE and Schneider who know the electromechanical side. That layer was missing," adds Jose.

Flutura wants to flip the usual model where data scientists try to deconstruct and solve engineering problems. It’s building a product for engineers to navigate the AI system and find solutions. The Engineers’ Workbench will enable engineers to apply AI without being experts in data science.

“We used to train data science folks in finance, retail, and supply chain," says Srikanth Muralidhara, Flutura co-founder and chief customer officer. “I have concluded that when it comes to heavy engineering, it is next to impossible to train data scientists on the engineering aspects. So Flutura is going to bridge this gap by making data science accessible to engineers."

Last year, Cerebra scored the highest for customer satisfaction in Gartner’s Magic Quadrant for industrial AI/IoT, beating products from the likes of GE and Siemens apart from better-funded US .

Still, the Bengaluru startup has to fight to win deals over rivals like AspenTech, C3, Arundo, and SparkCognition which enjoy a perception advantage by virtue of being from the US. Flutura CEO Raman recently shifted to the company’s registered headquarters in Houston to be closer to the business action.

The Flutura founders have come a long way after leaving IT services company Mindtree, where they steered a project to set the digital architecture for the Aadhaar programme. They were no strangers to big data. The hard learning curve was in coming out of the IT services mindset and spreading their wings as entrepreneurs with a global software product.

Flutura means butterfly in Albanian, but what the seven-year-old startup does is the power-lifting to move giants like Shell, Henkel and Hitachi to Industry 4.0. That transformation can be no less dramatic than seeing a pupa turn into a butterfly.

Sumit Chakraberty is a contributing editor with Mint. Write to him at chakraberty@gmail.com

Close