How Eni is enlisting IBM's Watson in the hunt for black gold

The Italian energy giant is creating a new business unit to explore how IBM's cognitive computing could help it find new oil reserves and make better business decisions.
Written by Raffaele Mastrolonardo, Contributor
The Jeopardy! winning Watson system. Image: IBM

Defeating two Jeopardy! all-time champions is no easy task. Finding new oil reserves is likely to be far harder, but it could lead to a payout that dwarfs the quiz show's prize pot — the reason why Italy's biggest energy company has turned to IBM's Watson for help.

Eni, which prides itself on its use of IT, recently announced that it will create a special unit within the organisation called W@Eni, which will be tasked with finding out whether, and how, the IBM computer system could improve its business.

For the oil giant, whose 2012 revenues came in at €128bn, using Watson means placing a bet on a hitherto unproven technology. However, according to its executives, it's a bet that might help it keep pace with its competitors and fits with the company's plans to give IT a more central position in its strategic thinking.

For the inside story on the creation of Watson, read: IBM Watson: The inside story of how the Jeopardy-winning supercomputer was born, and what it wants to do next

"We want to become a digital enterprise and we want to exploit the opportunities provided by big data. But the complexity and diversity of our business require tools that further advance technological capabilities. I think we found the answer in cognitive computing and in its Watson embodiment," Gianluigi Castelli, Eni's CIO, told ZDNet.

For IBM, which has just taken the first steps on the path it hopes will lead to the commercialisation of its creation, getting such a big name onboard is a promising development and one that might help win other customers over in future. "There are other Italian companies from the energy, finance and healthcare sectors that have showed interest in Watson. They all operate in markets where big data is a new competitive resource for their business," Nicola Ciniero, CEO of IBM Italia told ZDNet.

A new kind of team

According to Eni, the experiments with Watson are scheduled begin by the end of May, with the company currently putting together a 30-person team in charge of the task. "We are setting up a heterogeneous unit. We will bring in specific skills for every area where we want to apply Watson, then we will involve employees from the various business lines as we try to bring together highly-diversified computing skills," Eni's Castelli said.

The goal is to launch a series of pilot projects by the end of the summer, which should form the basis for some larger-scale initiatives that, ideally, will begin by the first quarter of 2015.

All in all, the investment in the pilot phase is set to be around €4m, which includes training, infrastructure, internal resources, and external support.

Estimated costs, though, could vary since cognitive computing is still largely uncharted territory. "This is a highly experimental activity for which we don't have any previous evidence to make precise estimates. We will adjust the investment on the fly, based on what we learn and the results we will get along the way," the CIO added. A team of five or six IBM staffers will help in the initial phase, though Castelli did not disclose any detail about the collaboration between the two companies.

What we know, though, is that the Italian corporation is looking to put Watson to work in those areas of its business where the expected return — in terms of new revenues, cost reduction and customer satisfaction — is deemed highest.

"Support in new oil fields exploration seems to me the obvious choice. But we are looking also at the trading business, where the risk of taking decision that are only partially informed is very high," Castelli said. What Eni is hoping for, according to its IT chief, is deeper and broader analysis coupled with speedier and more accurate decisions which should translate into better customer support, reduced exploration costs, and minimised risk in trading.

For its part, IBM hopes it won't be long before it can demonstrate how companies like Eni can put its cognitive computing services to work in the real world. Successful case studies might accelerate the adoption of the system among companies of all sizes. "Just like in the United States, where a $100m ad hoc program has been launched, I'm sure that in Italy too we will see the birth of a startup ecosystem that can lead to a new generation of apps that exploit cognitive computing," IBM's Ciniero said.

No quick fix

As much as Eni's management has faith in Watson's power to deliver benefits for the company, they're not expecting immediate results, acknowledging that it might take some time before cognitive computing brings dividends and that there are many unknown variables in the process, particularly when the time dimension is taken into account.

"How long will it take and which skills will be required take to define the ontologies, that is, the knowledge structure for each domain in which you want to put cognitive computing at use?" Castelli asked, adding that the answer is still uncertain even to Watson's creators. "The team that has developed Watson and applied it to Jeopardy!, cancer research, and in the medical insurance sector is very cautious with timelines. So I don't expect quick fixes: I would say it's definitely not a matter of a few months, but, at the same time, it won't take five years."

Caveats notwithstanding, Eni's hopes for using cognitive computing for better decision making are high. It's about "breaking free from algorithmic schemes where everything has to be foreseen and programmed, [and] entering an environment where communication between humans and machines is done in natural language and knowledge is enriched by specific training and by the interaction with humans," Eni's CIO said.

And then of course there's that little dream of finding some black gold where nobody else has yet. Can really Watson deliver there? "It's too early to say, but I'm persuaded that it will allow us to do it in shorter times and at a lower cost — provided that we're good in training it," Castelli concluded.

Read more on Watson

Editorial standards