Accenture Launches Specialised Services to Help Companies Customise and Manage Foundation Models

Accenture unveiled a comprehensive set of new services designed to help companies customise and scale the value of generative AI.

“Generative AI is accelerating reinvention and delivering value to our clients’ businesses faster,” said Julie Sweet, chair and CEO of Accenture. “In order to move from experimentation to scaled implementation, our clients need a strong data and AI foundation, which is underpinned by large language models that are tailored to meet their specific business needs, along with the training needed for their people to use and manage those models. Our new foundation model customisation services help our clients accelerate access and contextualise enterprise data to drive tangible value from generative AI.”

This is consistent with Accenture’s survey: 74% of the C-suite plan to increase their AI-related spending in 2024, up from 50% in the previous year, as companies move from proof of concept to customising foundation models with proprietary data.

This new set of Accenture gen AI services includes a proprietary gen AI model “switchboard,” customisation techniques, model managed services and specialised training programs.

Proprietary “switchboard” allows a user to select a combination of models to address the business context or based on factors, such as cost or accuracy. For example, one of the largest entertainment companies is currently testing the switchboard to compare how the same prompt would be interpreted by different models and how they perform before deciding on which to use.

Customise LLMs for specific business context, data sources and various techniques, along with managed services for ongoing finetuning and prompt engineering. Skills and capabilities to help clients effectively use and manage LLMs, including comprehensive training and certification programs. In addition, as demand for new AI skills has emerged, Accenture continues to collaborate with leading academic institutions. As part of its collaboration with the Stanford Institute for Human-Centered AI, Accenture has created its Foundation Model Scholar Program to provide certifications related to large-language model skills.

For example, many insurance industry employees review a lot of documentation to serve policy holders, which Accenture expects to benefit from generative AI, and illustrates what The Hartford is aiming to address in its collaboration with Accenture.

“We are pleased to collaborate with Accenture to help fully leverage the power of generative AI as part of our aggressive technology agenda,” said Deepa Soni, chief information officer, The Hartford. “With AI already mainstream at The Hartford, we understand its potential and will harness this power in conjunction with responsible AI principles. Our advanced architecture and ability to customise generative AI models will enhance our employees’ work across the value chain, further positioning us to serve our customers, agents and shareholders.”

Building on its $3 billion investment in AI, Accenture’s foundation model services and its proprietary LLM assets were developed by its Center for Advanced AI. Accenture has more than 1,450 pending and issued patents in AI, including proprietary capabilities in generative AI, such as fine-tuning, prompt engineering and model architecture.

“Companies that want to harness the power of generative AI will need an array of specialised models,” added Karthik Narain, group chief executive, Accenture Technology. “Our switchboard is designed to enable companies to do this at scale while managing costs and carbon footprint, aligning the right models to the business context. Our set of customisation services can help companies contextualise AI models for their unique needs, transforming every link in their value chain.”