With more artificial intelligence options to choose from than ever before, there are two main risks: an oversaturated market and a customer base that is almost too spoilt for choice.
Your favorite businesses are building their AI strategies. Every tech company has its own AI product or service. You either have your line manager in your ear, telling you to work out the best option for your company, or you’re sitting with the big leagues negotiating with your stakeholders over whether investing in OpenAI, NVIDIA, or Anthropic is best. In fact, 83% of our developer survey stated some support from their employer for adopting AI.
However, the near abundance of choice does not have to be a negative. In fact, the myriad of options now presents businesses with the solution to a pressing problem: being stuck with just one AI tool or model.
Why Is This a Problem?
The AI race is a busy one, both on the vendor side and the customer side. Vendors with B2B or B2C AI solutions are in constant battle with one another to create the tools most suitable for their customers. Over on the customer side, the pressure behind making the right choice is mounting, and for some, it is the difference between excelling this year or crashing all the way down into irrelevance.
For customers, making the right choice when it comes to their chosen AI model or tool is just the first step. They have to ensure that the model or tool works for their specific needs and use cases. It has to be affordable, customizable, and applicable. It has to be a tool that makes it easy for that business to discern ROI, and one that is both scalable and simple to manage.
Once that choice has been made, the hard work starts. A business might go months or even years with this one AI solution, and it could work really well, ticking all the necessary boxes. However, somewhere down the line, the next exemplary model comes out. It’s more powerful than any model before it, it has more applicable use cases, and it may even offer a cheaper solution. By then, the business has committed, but it might be considering either a transfer or an addition.
Even Salesforce’s CEO, Marc Benioff, has been open about switching from OpenAI to Google’s Gemini 3, indicating that business leaders need to adapt to flexibility.
So, how do they do that?
Don’t Get Locked In
In this scenario, the business begins looking into the new model, and it is impressed enough to want to give it a try. However, there’s a problem: it’s suddenly up against multi-thousand-dollar exit fees, transfer fees, and conversion costs. By this point, the business weighs up the pros and cons. Could making the move be worth it? Will adding this model to the tech stack soften the blow? What about the financial, data, and adoption considerations?
“There is a constant sense that a better model is always just around the corner, and being locked into a single vendor or model for years can quickly become a disadvantage,” Robert Sösemann, Senior Principal AI Architect, told SF Ben. “This applies just as much to businesses as it does to developers.”
“AI right now is a race between model providers and tooling platforms. Locking into one vendor too early risks falling behind. At the same time, companies need structure: a clear policy that allows teams to propose and adopt better models through an approval process, while preventing people from bypassing governance by using personal or unofficial accounts.”
Matt Pieper, a longstanding voice in the Salesforce ecosystem, said that the risk of being locked into multi-year contracts or agreements is not new, but it should always be considered when it comes to AI.
“Locking into any vendor has always had risks, especially when dealing with novel tech or first movers,” he told SF Ben. “The calculus is still the same with AI that it’s been with software for decades: multi-year contracts save money but limit your flexibility; the tradeoff is risk versus cost.”
“Personally, I think businesses should use the right model that is right for the task at hand. We’re only three years in, and models are becoming cheaper to develop and train, with a focus on speciality models. So each business should ask: ‘What’s our appetite for risk, and what are we looking to accomplish?’”
The Security Risks
If you’re a business that has decided not to pledge loyalty to just one AI tool or model, then there are, of course, a lot of benefits. On the flip side, the biggest negative impact or consideration is the security risks, especially if your business handles a sizeable amount of data.
“Flexibility does introduce real security risks,” Robert said. “People tend to put everything into AI systems, whether it’s source code, internal documents, or sensitive business data. Without clear rules, that quickly becomes dangerous.
“The biggest challenge is speed. AI models evolve so quickly that companies often can’t define or update their policies fast enough. Meanwhile, employees hear about new models through social media or consumer tools and want to try them immediately. That gap creates a high risk of shadow AI usage.”
Matt’s view on the matter is simple: every and any model you choose will come with its own set of security risks, even if the companies behind the models have taken steps to ensure data is being protected.
“There are countless stories of technology companies violating their terms of service to get a leg up on their competition, and we shouldn’t expect anything different with AI,” he said. “It’s once again a calculated risk.”
This is not to say switching between models or taking on a multimodal approach is inherently negative. Matt advises that if businesses want to make use of this kind of approach, it’s vital that they understand where their data is coming from and going to. Without a dedicated security team or robust processes, this can become a daunting task very quickly, but it is something that undoubtedly needs to be considered.
I Still Want to Use Just One AI… What Do I Do?
Here arrives the ultimatum. At this point, you might still be thinking: “I still want to stick with just one AI model or tool.” Is it sensible and feasible? It certainly can be.
“The main benefit of locking into a single model is simplicity: fewer security concerns, clearer policies, and less friction for governance and compliance teams,” Robert said. “From a risk management perspective, that can feel reassuring.
“In the long run, companies may stabilize on a provider or a model family while continuing to evolve the tooling around it. But we’re not there yet. For now, using a platform that abstracts model choice and continuously upgrades models on your behalf, while keeping pricing stable, is a very pragmatic approach.”
Matt’s sentiments echo Robert’s when it comes to simplicity, especially when considering whether model switching or adding on is contributing to ROI.
“Just like with code, you start small, learn the fundamentals, build up your expertise, and then layer in things like frameworks,” he said. “Model chasing can cause churn and noise, and can cause you to ask: ‘Am I spending more time maintaining than the benefits provided?’”
Ultimately, the decisions behind the number of AI models you choose to use depend heavily on risk versus reward. If you’re new to artificial intelligence or perhaps are a smaller business, then opting for an all-in-one solution might be best. Salesforce and Agentforce provide this for a growing number of businesses, especially if they have resources to spend.
However, having more than one AI on hand also has its use cases, especially if you want to use different models for different business tasks. Utilizing OpenAI’s suite of tools alongside other models is an example of this, especially if you are chasing more flexibility.
Final Thoughts
Whether you believe being trapped in a single vendor’s ecosystem is a significant liability or not, the scope to choose exactly how your AI strategy plays out has never been bigger.
Competition is fierce, new models break through the market every year, and customers’ expectations are only increasing. If there is one aspect you take away from this post, let it be to really consider the risk versus reward of every decision you make surrounding this topic. There has never been a time more critical than now to make sure your business remains at the forefront of your market, and hopefully, this post helps you make those decisions.
