Allegra Driscoll, the chief technology officer of Bread Financial, has a few ground rules when engaging with software vendors about generative AI capabilities. No agreements beyond one year, given the rapid pace of change. There will be times when she’ll double-spend on two vendors with similar capabilities if the use case is a priority and Driscoll wants to see who is more likely to deliver.
And the core focus of her conversations with vendors is also evolving. In 2023, at the start of the generative AI boom, Driscoll would discuss a software provider’s AI roadmap, key milestones, and what an investment would look like for the provider of private-label and co-branded credit cards. But now, there’s far more focus around how platforms are designed, talks Driscoll describes as almost philosophical.
“The conversations are going a lot deeper into the architecture of the third-party solutions, where in the past, I’ve been more focused on the capacity, security, and data privacy,” says Driscoll.
Nice’s chief information officer, Hadas Reisbaum, says she plans to leave her core systems in place for software that’s deeply embedded in the customer relations management software provider’s infrastructure. But she would like to see vendors evolve their pricing models and move away from the per-seat fee structure that’s most prevalent across the software-as-a-service industry.
“I think the clock is ticking,” says Reisbaum, who anticipates that bigger pricing structure changes could occur within the next two to three quarters. “It will become more outcome-based,” she added, meaning organizations like Nice will pay for service based on measurable results.
Even the buzzy, smaller AI startups may not be spared. Bread Financial works with upstarts like legal AI firm Harvey and AI content platform Jasper. But Driscoll says she could replace those vendor offerings as Bread Financial continues to develop its own AI platform.
Charles Guillemet, CTO of cybersecurity firm Ledger, says that it could be theoretically possible to rebuild the business software Workday does, but it would require far more effort than it’s worth. “If another company disrupts them with AI, we might consider moving away,” says Guillemet, especially if the alternative is cheaper and offers a stronger performance. “But for now, there’s no reason to move.”
He sees two paths forward: the first is that the large language model makers, like OpenAI and Anthropic, are able to pour so many resources into developing their product offerings that compete with SaaS that it becomes nearly impossible for anyone else to compete. But the second, which Guillemet favors, is that the technology advancements from AI hyperscalers will plateau and that competition will shift toward optimizing the cost of delivering software.
But the reality, according to Balazs, is that enterprises are discovering it is quite hard to get these unique SaaS-created agents to work together. He advocates for a more collaborative approach. “We want to make them expose their tools and skills, which for lack of a better word, is a new way of saying their API,” says Balazs. “It’s basically an AI API.”
Sagnik Nandy, the CTO at electronic-signature company Docusign, says he fields countless pitches from vendors but says his priorities are “dollar, people, time,” in that order. First, Nandy wants to know the upfront costs to sign a contract. From there, he asks questions about how many IT professionals are needed to implement a solution (vendors always provide a low estimate, Nandy says) and then seeks to understand how much time is needed before value can be unlocked and measured.
Nandy says he’s especially wary of vendor pitches that may generate value for his team, but where shifts in processes could create more work elsewhere.
“A common pattern I sometimes see is that the CTO might get value, but the CIO’s work goes up,” says Nandy. “I don’t go for those kinds of pitches.”
John Kell
