That is how Lori Mitchell-Keller describes Google Cloud, the search giant's cloud computing business, in its approach to working with companies in a variety of vertical markets including retail, banking, healthcare, manufacturing and many others.
"Every company in the world has deep domain expertise about what they are producing," said Mitchell-Keller in an interview with ZDNet recently. "We can make them more efficient to use their human talents in areas that are their differentiation."
Mitchell-Keller, who holds the title global leader and vice president of industry solutions for Google Cloud, came to the company a year and a half ago after thirteen years at enterprise software giant SAP AG, where she ran global industry solutions.
The infrastructure business in all vertical markets has been a vibrant field of competition for Google Cloud and its competitors, Microsoft Azure and Amazon AWS, as well as the enterprise technology titans such as Oracle, SAP, IBM and others.
But a change has been taking place at Google Cloud, says Mitchell-Keller, a shift from a pure infrastructure approach to a "much more strategic process" in working with industries.
"Products such as visual search, visual inspection, call center AI — these are not things that are produced overnight," said Mitchell-Keller regarding technologies Google Cloud had in the works before she arrived.
"The change now is that we are much more focused on what are the customer problems that we are seeing as we talk to big-name customers," she said, noting that Google Cloud has "eight of the top ten customers in almost every single vertical."
That customer base means Google has been getting "inflow […[ in terms of what are customers struggling with."
"Each industry is different, each customers is different, each one's context is different," observed Lisa O'Malley, the senior director for product management for the industry solutions group, who joined Mitchell-Keller in the same interview with ZDNet.
O'Malley came to Google Cloud a little over a year ago from PayPal where she ran enterprise solutions.
"We really dive into those challenges that they see, the opportunities that are in front of them, and deeply understand the outcomes they care about; we don't want to just automate things."
Take the example of retail. The problem of "cart abandonment," people putting something into an order basket on a Web site and then never going to check out, is not the entire story.
"We don't think it's really cart abandonment, it's search abandonment: people never really get anything in their cart," says Mitchell-Keller.
One of Google's customers for enhanced search is Macy's. The company last year instituted Google Cloud's Retail Search program. When people search for items, they may actually get appropriate results where Macy's prior search would mistake the intent.
A search on the phrase "dog costume" will now yield costumes you can put on your dog, versus results under the previous system that would show apparel for people with dog images printed on it.
"No matter what the intent of the shopper is, we want to surface that vast assortment that we have," Matt Baer, the chief digital officer for Macy's, told ZDNet. "That's what Google Cloud Retail Search helped us to do."
That includes handling daily updates to Macy's catalog of inventory, he said.
"If you look at the long tail of search results, and what would have otherwise been a very manual process of improving them, that is not something that we would have ever gotten to," said Baer. "With Google Cloud Retail Search, we don't have to worry about fixing what would have been perceived as a broken experience."
A similar application is IKEA's use of recommendations capabilities, part of Retail Search, to anticipate related items that shoppers may be looking for. That application of the technology has boosted the "click-through" rate for IKEA five-fold, said Mitchell-Keller.
Much of being a tool for companies comes down to understanding the data they posses, and how it can be most effectively used.
"Companies in every industry are sitting on a ton of data, that's not news," said Mitchell-Keller. "But every industry is really different in terms of how they are approaching that data.
Finding more effective ways to use data, including reducing the amount of data needed, is a key concern.
An example is visual inspection for manufacturing, an offering that Google Cloud made live this past June, to help companies that make things do quality checks of products as they "come off the line."
The goal is to reduce the cost incurred by quality issues that can consume up to 15% to 20% of a manufacturer's revenue.
"If you have a visual inspection tool today, you have to train it with so many images, it is a pretty intensive process," observed O'Malley. With the Google approach, the total amount of training images can be reduced by three hundred times, she said. Some of that comes from the years of work that Alphabet had already done training image recognition programs.
The reduction in training data is "a huge overhead savings for customers," said O'Malley. Among companies that have benefitted from the smarter approach to inspection is contract assembly giant Foxconn, one of the assemblers of Apple's iPhone. Accuracy was increased by an order of magnitude in inspections of phone assembly, said Mitchell-Keller.
Google has taken three approaches to serving industries, said O'Malley.
First, Google Cloud has "productized industry solutions," turning pieces of code into things that work "out of the box" to bring immediate benefit.
"If you are really building the solution for a problem, it shouldn't take a busload of people to implement it," said Mitchell-Keller. That is one of the criteria for what she calls the "litmus test" for cloud services. "The cost shouldn't dwarf the problem."
Second, the company has what are called "packaged solutions" that come with things such as "prebuilt connectors." One such program is "Doc AI," which ingests documents such as drivers licenses or mortgage agreements. That program was adapted for the lending business.
"We're taking horizontal components, putting them together, and then adding industry capabilities on top," said Mitchell-Keller.
Perhaps most important, said O'Malley, is the company's expanding number of partnerships with firms having industry expertise. "We can't do everything ourselves," said O'Malley. "We want customers to be able to use partners' solutions as easily as they can use our solutions with their data."
All three aspects are part of moving beyond a strict engineering fixation. Part of the change in leadership from Diane Greene, who ran the business in 2017 to 2018, to Thomas Kurian, who runs it now, is "the change in terms of a very engineering-oriented culture to a culture about using engineering to delight customers."
The most important benefit should be not just solving current problems such as click-through rate or document understanding, but also seeing down the road to what challenges lie ahead.
"What do we think, on an ongoing basis, is going to happen over the next two to three years" in a given industry, is how Mitchell-Keller describes it.
"And what do we think the big business problems are going to be that customers are going to face, and how do we get ahead of the curve and solve those for customers before they even get there."
An important broad challenge goes back to the issue of data, and that is the end of cookies on the Web as traditionally used, and the need to use first-party data instead.
"Being able to do personalized recommendations has been the Holy grail for a while" in many industries, observed Mitchell-Keller. "Our customers have a lot of data about their customers," and how to organize and analyze that data is a large quest.
Mitchell-Keller gives the example of Albertsons, the grocery store chain. "They have pharmacy information, lots of regulations around that, and then they have grocery information," she noted. The challenge, she said, is "how do they create that master [data record]" for each customer.
Google Cloud can approach the problem by borrowing from approaches applied in other areas. "We have the opportunity to use a lot of the things we've done, like the healthcare data engine, to then help a retailer in terms of doing that, creating that master data record," she said.
The data in such cases continues to be in the possession of the customer, said O'Malley. "They hold the data, we just enable the processing on that data," she said.
Another future challenge, said O'Malley, are the many ways that industries will reach customers, what can be broadly lumped together as "omni-channel."
"As we come out of the pandemic, nobody feels everyone is going back to the way it was, and so shopping patterns have changed," she said. Only it's not just shopping, it's other things, too, such as banking. "How do we help financial institutions that are used to banks on every corner think about a world where that online and offline customer journey has to be woven together very seamlessly?" is how O'Malley frames the challenge.
The current global supply chain crunch may be a model for long-term industry approaches, said Mitchell-Keller. "A lot of things we are pioneering in the supply chain are going to be in high demand as we come out of the pandemic." That includes demand forecasting, and a project under development at Google Cloud to predict where inventory should be. "We have solutions we are working on with the Geo team around fleet routing and last-mile delivery, and how do you do that more efficiently."
Another project early in development, currently in a preview form, is a digital twin for the supply chain. Digital twins is a common method these days in the industrial and manufacturing worlds, modeling a kind of mirror image on the computer of a real-world object, such as equipment, and then being able to probe and simulate interactions on the computer model before taking action in the real world.
In Google's case, the twin "allows you to identify bottlenecks, identify where things are going wrong, particularly with severe weather," said Mitchell-Keller. "I think that supply chain space is going to get a lot of focus, and the things we're working on are going to be critical for our customers going forward.
Of course, with so much chatter about digital transformation, and digitization, there is a danger that customers may want to simply check the boxes to have a "solution," including a chat bot or conversational AI, without knowing what they really need.
How does Google Cloud shift the discussion to the quality of work?
"Part of the challenge we have is that a lot of our customers hear digital transformation and don't necessarily know what that means," conceded Mitchell-Keller. "That's why the cultural change at Google Cloud is important," she added.
"When I'm building the industry solutions team, I have not been hiring software or technology people, I have been hiring people that have decades of experience in these particular industries," she said.
The person running financials, she pointed out, is a former EVP at Wells Fargo. Another executive, running retail, was the chief digital officer at Neiman Marcus. And the head of consumer packaged goods efforts at Google Cloud was president of personal care products at Kimberly-Clark.
"They know what challenges and problems those industries are dealing with," said Mitchell-Keller. "They can surround themselves with the engineering teams to figure out how to solve those problems."
With so much industry knowledge being assembled by the Google Cloud team, an existential question arises for Google's customers: What is their core competency? Is Google Cloud at some point better than a financial firm at financial services? Or better than a packaged goods company at packaged goods design and marketing?
To Mitchell-Keller, it comes back to being a "toolset," as she terms the Google Cloud resources.
"Every company in the world has deep domain expertise about what they are producing, whether its Merck with pharmaceuticals, or J&J with personal products, or Deutsche Bank in terms of how they do their banking and manage their portfolio — we don't do that," said Mitchell-Keller.
"We don't have the deep domain expertise about the piece that they are so good at doing," she said. Google can take over the mundane, she offered. A bank, for example, may not need to spend teams of its own people on fraud detection, for example.
The other things, the higher-level functions of industry, she said, remain the province of the customer.
"There are not any capabilities that we would have to be able to actually run that business from the ground up."