Product or feature? A key AI debate could leave CIO strategies in limbo

The rapidly evolving AI ecosystem, where new products and services seem to appear daily, presents CIOs and IT purchasing leaders with increasingly challenging decisions, in part because of uncertainty about where the AI market may ultimately be headed.

One major debate, with implications for CIOs and IT buyers, is whether AI will primarily be a product or a feature after the AI market sorts itself out.

On the one side are AI pure-plays offering niche and often task- or industry-specific point solutions, as well as AI generalists such as OpenAI, Meta, and Google, whose standalone large language models (LLMs) can be integrated with other IT systems. On the other side, IT vendors such as Salesforce, ServiceNow, and many cybersecurity providers are rapidly adding AI features to enhance and transform their core products.

A third option, where end-user companies build their own AI tools, seems to be seems to be imperiled, at least for now, by a huge majority of AI proof-of-concept projects failing.

Finding value from AI

The challenge for CIOs and other IT leaders is to determine what type of AI will best benefit their organization with so many options and possible paths currently available. Over the long term, many experts see the market consolidating, with a small number of LLM vendors or a small number of IT platform providers dominating the market.

CIOs and CAIOs looking at AI products need to understand what they’re buying and how they can get value from it, says Andreas Welsch, chief AI strategist at IT consulting firm Intelligence Briefing.

“Vendors are passing on the higher cost for gen AI and agentic AI to their customers and often justifying it with increased value,” he says. “AI decision-makers need to do a thorough ROI calculation to determine if the value is really there and on what level — transaction, task, process, role, or outcome.”

Welsch predicts the AI market will largely merge into the SaaS space, with many software vendors offering agentic AI as an added feature.

“Software vendors have been adding AI and agentic AI to their platforms to increase lock-in effects and fend off more nimble players and open-source alternatives,” he says. “By offering their own agentic capabilities, vendors can drive upsell and cross-sell programs and pitch a larger value of their 20-plus-year-old architectures.”

Easy to manage

A recent trend toward software vendors and platform providers including AI as a feature should make CIOs’ jobs easier because of the manageability and security benefits, says Steve Logan, CIO of field service software vendor Simpro.

Some standalone AI products seem to have struggled to find a market, prompting companies like Microsoft and Google to bundle AI tools into premium product suites, he notes.

“Over the past year, AI rapidly shifted from standalone tools and features buzz toward being primarily an integrated feature within core SaaS platforms,” Logan adds.

His focus, as a CIO, is on evaluating total bundled value, balancing innovation with integration and security, demanding ROI, and mitigating lock-in risks.

Over the long term, a lot of AI functionality will be delivered as an add-on feature to software packages or IT platforms, predicts John-David Lovelock, a research vice president and analyst at Gartner. In many cases, IT leaders won’t have a choice; the new feature will just show up.

Some IT providers are now charging extra for new AI features, but over the long run, the good news for CIOs on tight budgets is that these embedded features will likely show up without immediate price hikes for the base software or platform, Lovelock says. IT providers who aren’t AI first movers in their space will likely build AI into the cost of their products.

“There’s no doubt there will be a few holdouts that charge extra for the package, or that will try to do something with a consumptive piece to it,” he says. “But overwhelmingly, it’ll be embedded, and it’ll be free.”

In the interim, with pricing models for AI up the air, and new models emerging, CIOs will need to pay close attention to the bottom line when purchasing AI services.

A market for both types of AI

Other AI experts see a marketplace for both built-in AI and standalone products. Off-the-shelf application providers that use service dashboards will either add AI functionality or they will die, says Mark Townsend, co-founder and CTO at AcceleTrex, a provider of an expert referral service.

However, there will be some AI uses and data sets that require separate products, he adds. In some cases, AI tools created in-house by end-user organizations will partner with AI vendors that link them to data lakes and both internal and external services, he adds.

On one side are the providers of integrated AI functionality, including Salesforce and many cybersecurity vendors, Townsend says. On the other side are component vendors allowing companies to build their own AI tools.

Internal AI projects “often require several pieces, like Lego blocks, that need to be assembled and integrated,” he adds. “Teams can build their own AI applications from these components.”

AI will be both a product and a feature, depending on the needs of the users, adds Gulce Karsli-Rozenveld, CEO at online flight booking service Oojo.

“AI isn’t one or the other; it’s both,” she says. “Think electricity — it’s both a utility and a force powering other technologies. This duality powers its added value, with AI as a standalone environment and as an embedded layer enhancing existing systems.”

In the online travel industry, both chatbots and AI agents will be useful, she says, with the next generation of chatbots handling most bookings, and agents providing support or dealing with non-standard travel experiences. Tools like ChatGPT may become marketplaces where travel companies plug in.

The lesson for CIOs and other IT leaders is that they should remain flexible, Karsli-Rozenveld adds.

“The CIOs, CTOs, CAIOs, and even CEOs need to do their due diligence if they are betting good money on it,” she says. “But even more importantly — since the market is still evolving — I would say it is vital to stay on your feet and be ready to adapt.”

In the short term, many AI startups will flare up, then fail, she adds. “No tech is a 100% sure bet 100% of the time,” Karsli-Rozenveld says. “If tech leaders can see that a chosen AI platform, product, or feature is not operating as they would expect, there is still time to cut losses and find an alternative.”