Tips and tricks for negotiating with AI vendors

IT pros have a tough job, and one big stressor is the need to negotiate with third-party AI vendors.

Nonetheless, it’s becoming part of the IT team’s job to deal with vendors and advise higher-ups on how to choose the right service provider. That can be challenging, and often overwhelming, with questions about price, usage, vendor switching, and more.

To help with the process, it’s good to know what to prioritize for negotiations and what questions to ask—and not to ask.

AI isn’t a strictly “new” technology, said Jack Hogan, VP of advanced growth technologies at SHI International, but it’s increased in importance in recent years. That has meant more responsibility on the shoulders of IT professionals, who are often asked to give feedback on what vendors to work with.

“AI vendors that are out there range in where they’re adding value across the stack or the chain of how you can apply advanced AI capabilities—perceptive AI and now generative AI, moving into agentic AI,” Hogan told IT Brew.

Leverage questions. In the past year, there’s been a notable shift to AI integration affecting nearly every aspect of the traditional tech business, with executives referring to the technology as likely the biggest change the industry has seen. That’s forced organizations to become more selective with regard to vendors, which comes with its own complications.

“It’s a lot messier than it used to be traditionally,” Gartner VP Analyst Kjell Carlsson told IT Brew in August. “There’s an element of uncertainty that we don’t know how to do migrations in the same kind of way…We’re just not even sure what we need to migrate.”

Shafqat Islam, president at Optimizely, is the single largest buyer of AI at his company, he told IT Brew. The biggest question Islam encounters is the cost model and pricing levers; with a lack of predictability in the AI space, it’s difficult to find stable ground on how much technology costs on a month-to-month or annual basis.

“Understanding the pricing model, understanding what the guardrails are, what happens if you go over—the IT pro has to help guide and be a shepherd to the CMO through that,” Islam said. “Maybe ask for more predictability, or ask for constraints.”

With that type of chaotic approach, an outcome-based pricing model is of interest, in which customers pay per successful completed action (such as the AI resolving a support ticket). The most important thing is to make sure you’re approaching the negotiation with as much information as necessary.

“If you’re an IT buyer and you’re pushing for outcome-based pricing, you better at least loosely understand what the outcomes are that your business partner cares about,” Islam told IT Brew. “There’s a lot of leverage for the buyer and the IT guy who’s procuring that software.”

Power up. The power increase of AI models means that a capability assessment is essential when deciding who to go with, Hogan said. Meaningful improvements need oversight, which requires additional resources, and that has to be part of the equation during the negotiation process.

“You experiment with it and you create this demo that looks great and it does a bunch of things, then when you take it to the adoption phase, you start to see a bunch of challenges,” Hogan said. “It’s really at that point where you need to have a trusted partner that helps you recognize that the concept is great, but the implementation may have a lot of flaws, and it could have flaws in the form of business risk or even financial risk.”