5 tips for choosing the right AI model for your business

The growth of generative AI (gen AI) has been fueled by high-profile Large Language Models (LLMs) such as Open AI’s GPT-4o, Google’s Gemini, and Anthropic’s Claude.

But while these larger models grab the headlines, another model range has gained traction. Some experts believe small language models (SLMs) could be the future of genetic AI.

Also: Asana launches a no-code tool for designing AI agents – aka your new “teammates”

According to research firm Gartner, LLMs have traditionally dominated language model development, but SLMs offer potential solutions to key challenges identified by functional leaders, including budget constraints, data protection, privacy concerns, and risk mitigation related to AI. Business leaders may have to choose between larger and smaller models when exploring genetic AI.

So who will win the fight? Five business leaders tell us what they think.

1. Consider domain-specific opportunities

Claire Thompson, group chief data and analytics officer at financial services giant L&G, said she expects small and large models to have a place in business operations. However, she also believes that today’s high-profile models could be optimized for new use cases.

“I can imagine a situation where some of the LLMs could start to undertake further training on specific topics to gain more detailed insights, and I can imagine this happening more and more often,” she said.

Although there is a gap in domain-specific models, Thompson told ZDNET she’s not sure many companies would devote human and financial resources to internal development.

“I don’t know if you would build your own,” she said. “When I talk about building models, it’s more about leveraging existing models internally and leveraging your data in a secure environment to achieve results.”

Also: Technologist Bruce Schneier on security, society and why we need “public AI” models

Big or small, Thompson said, the future lies in domain-specific models.

“I think we’re going to start getting more tailored models,” she said. “For example, you could see how you could tailor a model to medical information, climate issues and ESG, and asset markets. It’s these specific use cases where you could release more tailored models.”

2. Choose the right horse for the course

Nick Woods, CIO of MAG Airports Group, is another digital trailblazer who said the future of Gen AI is likely to be a mix of large and small models.

“I don’t think there’s a one size fits all,” he said. “And I think which model you choose depends on the use case in your company.”

Woods told ZDNET it’s not unusual for experts to say the organization should start an AI program. His answer? “No, that’s the last thing we should do.”

Plus: Gartner’s 2025 Technology Trends show how your business needs to adapt – and quickly

Woods said leaders should focus on the business transformation agenda and decide which tools, including genetic AI, can help achieve the right results. “For example, we might want to run a small, specific model at the edge to solve a specific use case, such as detecting when an airlift is docked,” he said.

“I might be doing something different if I’m trying to build a model for a question like, ‘What does global air traffic look like and how will it respond to weather changes?'”

In short, Woods said, choosing a model is about choosing the right horse for the course.

“I think there are a lot of small, large-scale models being deployed at the edge for specific use cases,” he said. “It’s almost inevitable. However, I still believe that some large models will prevail.”

3. Consider the context

Gabriela Vogel, senior director analyst in Executive Leadership of Digital Business at Gartner, said her conversations with CIOs suggest that small, domain-specific models have an important role to play – at least in the short term.

“The clients I talk to are trying to find and create models that apply to a specific context,” she said. “They are not necessarily large, general models, but models that are tied to small databases specifically for a particular application.”

Also: Perplexity AI’s new tool makes researching the stock market “enjoyable.” Here’s how

Vogel told ZDNET that more and more companies are moving from exploration to production-generation AI services using SLMs.

“They are making this shift because they have tested a lot,” she said. “They’ve seen what works and what doesn’t work on larger models, and then they try to be more specific and apply that approach. I have personally seen this with my customers.”

4. Walk small to reduce hallucinations

Ollie Wildeman, head of customer experience at Big Bus Tours, said the choice between SLM and LLM depends on the use case – and for many companies the choice is likely to be smaller rather than larger.

He told ZDNET how Big Bus Tours uses Freshworks Customer Service Suite, an omnichannel support software that includes AI-powered chatbots and ticketing. The company also uses an AI-powered virtual assistant from Satisfi Labs that connects to its website and handles basic customer queries.

“Satisfi’s AI technology only sources data from the specific companies they work with,” he said. “The company’s technology is not connected to large AIs like ChatGPT or other tools – they do it themselves.”

Plus: Today’s AI ecosystem is unsustainable for almost everyone except Nvidia, warns a top scientist

Wildeman said this cohesive approach creates business benefits – executives can be confident their data is being used carefully to achieve results.

“This way your data is more secure because you know where it comes from and what processes it uses,” he said. “You’ll also have fewer hallucinations because you know the model you’re using is designed for the type of business you’re in.”

These results lead Wildeman to conclude that smaller, domain-specific models will be important for companies.

“I think for enterprises the choice of model will be more specific, while for the general user probably those huge free models that you see everywhere will be popular.”

5. Focus on your first-party data

Rahul Todkar, head of data and AI at Tripadvisor, said the right model for a business may not just be a question of big or small.

Professionals can try both models. However, Todkar told ZDNET that purpose-built and tailored models are the future of AI, whether defined as large or small.

“Take the example of Mistral 7B, which is a relatively small model compared to other LLMs, but performs excellently on certain tasks,” he said. “So for me the future lies in adaptable models.”

Plus: Anthropic’s latest AI model can use a computer just like you do – bugs and all

Todkar suggests that the key to AI success is ensuring that the model uses your data safely and effectively.

“It’s not about the training size or the features of the model, but rather it’s about taking that model and applying it in your context with your first-party data,” he said. “Then you can go beyond standard models and use the insights from your data. So the answer will be somewhere in the middle.”