Computer vision technology, chatbot reboot on tap for 2020

Computer vision becomes fashionable

That said, 2020 will see new deployments as well. Computer vision technology is one example, with retail expected to emerge as a key vertical market for such AI projects.

"Of all the AI technologies available out there, this is probably … among the top technologies that can completely transform the retail value chain," said Rajashree R., global head of retail strategic initiatives, products and innovation at Tata Consultancy Services.

Rajashree R.

Computer vision technology has wide applicability in retail's fashion segment, she said. Those areas include range planning, inventory management, personalization and returns. Range planning, for example, involves determining the styles, textiles, colors and other attributes of a fashion line. Designers and buyers, typically, would visit fashion shows to get a sense of what's to come and plan accordingly. But computer vision lets fashion retailers pull together the highlights from fashion shows and combine those with images from Instagram or Pinterest to help predict clothing trends.

Of all the AI technologies available out there, this is probably … among the top technologies that can completely transform the retail value chain.

"What the human eye cannot synthesize, computer vision can," Rajashree said.

Online retailers will be the early computer vision technology adopters, she said. Stores in the fast-fashion category, which aim to quickly capitalize on runway trends, also have some urgency to pursue computer vision, she added. Other types of retailers won't be following, however.

"I don't see adoption among traditional department stores and big-box retailers," Rajashree said.

Fashion merchandising, she added, remains more art than science for such retailers. So, they are therefore less likely to trade their traditional methods for AI.

Other retail segments likely to adopt computer vision include large grocery stores, where use cases will range from store operations to inventory management.

Chatbots get a talking to

Enterprises in industries such as financial services, retail and healthcare have rolled out chatbots in abundance, but the results appear mixed.

CGS, an outsourcing and business application provider based in New York, found most of the consumers it surveyed would rather get help from a human than a chatbot. Eighty-six percent of the more than 1,000 U.S. consumers polled in the company's Customer Service Chatbots & Channels Survey, published in October, said they "prefer to interact with a human agent" rather than AI when using a texting-based messaging service.

John Samuel, executive vice president at CGS, said that number reflects a couple of factors. Some chatbots are not consumer-friendly, he noted. He said those deployments may be based on corporate FAQs or knowledge bases that "are limited in what they provide to consumers."

John Samuel

A chatbot's underlying technology may also limit its helpfulness, Samuel added. Consumers in a rush for information become frustrated with chatbots that meander through a scripted algorithm to get to the point where they can provide assistance.

But, despite the challenges facing existing chatbots deployments, Samuel sees no lag in the technology's adoption.

"We don't believe there is going to be any reduction in driving forward with the technology," he said. "Probably, in 2020, we will see a good increase in the use of chatbots."

Samuel's optimism stems, in part, from the ongoing advancement of natural language processing and understanding.

"That technology is evolving every day," he said. "Just in 2019, the natural language capabilities of the cloud providers have evolved."

In addition, chatbots have an opportunity to improve through integration, Samuel noted. Linking chatbots with enterprises' CRM systems and data platforms will boost customer experience (CX).

Christine Livingston, chief strategist for AI at Perficient, a digital transformation consulting firm based in St. Louis, also noted the importance of integration. She pointed to the example of an HR chatbot that informs employees they have paid time off but doesn't specify how much. Such chatbots fail to "execute a meaningful task," she said. "They need to integrate into the system of record."

Christine Livingston

Samuel said the integration task will be easier for born-in-the-cloud companies with modern infrastructures that simplify data extraction. Older businesses may need more help.

"You are dealing with multiple platforms, data sprawl in multiple repositories in some cases," he said. "Legacy platforms are not easy to integrate."

SaaS vendors, such as ServiceNow and Salesforce, are incorporating chatbots into their platforms, however, easing the integration burden, Samuel noted.

Organizations with multiple chatbots face another integration issue: keeping their virtual assistants in sync. Livingston said Perficient worked with a car dealership to create a chatbot for car owners with questions about specific vehicles. The dealership eventually discovered that other parts of the organization -- sales and finance -- had also unleashed chatbots. As a result, customers could find themselves getting different answers from the different chatbots. The dealership is now looking at consolidating a platform for more unified CX, Livingston noted.

"A lot of times, customers recognize how easy it is to do a very simple use case," she said. "It's more challenging to create a comprehensive, AI-enabled customer engagement strategy."

Livingston said she expects to see earlier adopters of chatbot technology to pursue platform consolidation in 2020 "as they progress to the next maturity phase."

Clouds seek new levels

Cloud computing, now over a decade old, still has some maturing to do. Industry executives believe enterprises will pursue more sophisticated migration and application modernization approaches.

John Gray, CTO at InterVision, an IT service provider based in Santa Clara, Calif., and St. Louis, said fewer organizations plan to pursue lift-and-shift cloud migrations in which applications move to the cloud without modification. Enterprises are interested in retooling apps as microservices, running them in containers and preparing for serverless architectures. Those aims compel IT managers to consider refactoring earlier in their cloud migration plans.

John Gray

"More and more IT teams will refactor their applications before going to the cloud, rather than considering this process after they've already moved to the cloud," Gray said. "We are now starting to see … a more proactive approach to refactoring."

But some customers may move beyond refactoring. Dave Sellers, general manager of multi-cloud at World Wide Technology, a technology solution provider based in St. Louis, said customers may decide to rebuild applications in the cloud rather than refactor them. He said customers make such decisions on a case-by-case basis, suggesting advances in technology could influence customers to rewrite applications.

"With cloud service offerings becoming more valuable every month and with the performance of serverless offerings getting continuously better, it can be compelling to entirely rewrite an application in the cloud to leverage these benefits," Sellers said. The rewrite-in-the-cloud strategy can also decrease the overall cost of ongoing application maintenance, he added.

Dave Sellers

Sellers said the trend in 2020 will be toward cloud-native development but didn't hazard a guess as to what percentage of customer apps would follow that path.

"I do think there will be a lot more apps built in the cloud next year than in the last couple of years," he said.

Multi-cloud deployments, particularly the multiple-public-cloud variety, will also gain ground in 2020, as customers become more analytical when it comes to placing workloads.

Multi-cloud arrangements spanning private cloud, a colocation facility (see sidebar) or a single public cloud are fairly common. But customers running applications across multiple public clouds have been rarer.

Colocation market set for growth

The colocation services market is poised for growth in 2020 and beyond.

IHS Markit, a market research firm based in London, predicted the global colocation services market will reach $38.8 billion in 2023, growing at a compound annual growth rate of 8.4%. One factor behind the growth: Organizations are factoring in colocation facilities as part of their multi-cloud strategies.

Ben Niernberg

The inclusion of colocation centers can help customers keep their public cloud data transfer costs in check. Getting data into and out of the cloud can prove expensive, noted Ben Niernberg, executive vice president of sales and services at MNJ Technologies, a technology and managed service provider based in Buffalo Grove, Ill. The issue is compounded for enterprises pursing data analytics on one of the hyperscale platforms -- those organizations have more and more data coming out of the cloud.

That's were colocation comes in. Colocation facilities offer private linkages to the public clouds that can reduce data transfer costs compared with public internet connections.

"Colocation facilities have direct connections, or express routes, to public clouds," Niernberg noted. MNJ Technologies offers colocation services and also partners with third-party providers, such as Equinix.

"It's a huge part of our business," Niernberg said. "We are expecting 30% to 40% growth in that area."

Sellers said customers who have been in the cloud for a while and have found success optimizing on a single cloud vendor will be in a position in 2020 to pursue a "best-of-breed scenario." A business, for example, might run analytics on Google Cloud Platform, Office 365 on Microsoft Azure and machine learning on AWS, he explained.

Jeff Ton

"As each one of these clouds continues to release more services, they tend to leapfrog each other in some technologies," noted Jeff Ton, senior vice president of product development and strategic alliances at InterVision. "You are starting to see this becoming part of a company's cloud strategy: 'We are going to put our AI models on this cloud [because] they do a great job with machine learning and AI.'"

Early multi-cloud use has been somewhat accidental as individual developers spun up cloud instances on AWS, Azure or Google, Ton said. But enterprise thinking is becoming more strategic, he noted. "Let's leverage the cloud for what [vendors] are best at" is how he summarized the emerging philosophy.