Specialised AI models set to outpace general-purpose LLMs by 2027, says Gartner

Developing task-specific AI models often involves techniques like retrieval-augmented generation and fine-tuning, making enterprise data a key differentiator that demands robust management, quality control, structuring, versioning, and governance

A new forecast from Gartner, Inc. suggests a major shift in enterprise AI strategy, projecting that by 2027, small, task-specific artificial intelligence (AI) models will be used at least three times more frequently than general-purpose large language models (LLMs). This growing preference reflects an industry-wide move toward precision, speed, and efficiency in AI deployments.

Although general-purpose LLMs such as ChatGPT or Claude are capable of handling a wide range of queries, they often lack the depth needed for specialised business scenarios. In contrast, smaller models tailored to specific tasks or domains are proving more reliable and cost-effective, delivering faster results while consuming significantly less computational power.

“The diversity of enterprise workflows and the demand for high response accuracy are accelerating the adoption of purpose-built AI models,” said Sumit Agarwal, Vice President Analyst at Gartner. “These models reduce operational overhead and offer better alignment with domain-specific goals.”

The development of such models often involves techniques like retrieval-augmented generation (RAG) and fine-tuning on proprietary data. As a result, enterprise data becomes a critical differentiator, requiring strong data management practices—ranging from quality control and structuring to versioning and governance.

Turning AI into revenue

Interestingly, Gartner predicts a shift in mindset from protecting proprietary AI assets to monetising them. Companies may begin licensing access to their fine-tuned models, offering them as products to clients, partners, or even competitors. This could mark the beginning of a more open, collaborative AI economy.

To prepare for this transition, Gartner recommends organisations start by piloting small, contextualised AI models where general LLMs underperform. Businesses should also consider a composite model strategy—integrating multiple models within structured workflows to enhance accuracy and resilience.

Equally important is talent development. Enterprises must invest in upskilling teams across AI engineering, data science, compliance, and business operations to build and sustain domain-specific AI initiatives.

As businesses move beyond one-size-fits-all solutions, the next wave of AI innovation appears to be smaller, sharper, and more strategically aligned with real-world tasks.