Cognizant is rolling out a set of new offerings that will make it easier for clients to apply AI in their business operations.

Some of the new offerings are professional services. Others are software products, such as language models optimized for vertical-specific automation use cases. Both solution collections use Nvidia technology under the hood. 

One vertical on which Cognizant is placing particular emphasis with the new offerings is the industrial sector. The company says that it will help manufacturers create digital twins of their plants to find ways of boosting operational efficiency. A digital twin is a simulation that is regularly updated with fresh data about the system it’s designed to emulate, which makes the simulation more accurate. Cognizant plans to power its digital twins using Nvidia’s Omniverse software kit, which includes rendering tools for generating simulation graphics, AI models that can enhance those graphics and related components.

“Digital twins enable manufacturing and warehouse managers to model changes without disrupting operations,” Vibha Rustagi, the Global Head of IoT & Engineering at Cognizant, told Boardroom Insight. “Different floor configurations can be modelled and tested for effectiveness before they’re implemented, ensuring that equipment, storage and workstations are optimally spaced to reduce bottlenecks and waste. Additionally, simulated workflows can provide insights into how materials, people and products move through the plant, revealing potential inefficiencies and unnecessary steps.”

The digital twins that Cognizant puts together for industrial clients integrate data from multiple sources, Rustagi explained, including records stored in legacy systems. Increasing the amount of information in a digital twin makes it more useful for decision-making. But there’s also a trade-off: the more production line data users have to review, the more difficult it is to run analyses. That’s where AI comes into the picture.

AI models “can enhance the value of digital twins by helping to identify patterns from data and sensors and suggesting changes that might not be apparent to humans,” Rustagi detailed. “Plant managers can optimize for different variables, such as cycle time, energy usage or consumption of other resources.”

Besides optimizing the configuration of production lines, AI also lends itself to monitoring those production lines’ KPIs. In that way, Cognizant sees its AI-infused digital twins helping clients “ensure that efficiency and resilience are built into operations,” Rustagi said.

The new AI offerings that Cognizant is rolling out also cover several other use cases, most notably AI agent development. The company plans to provide clients with the technology building blocks needed to build such agents, starting with language models. 

Cognizant says that it’s developing LLMs optimized for industry-specific use cases. According to the company, one fine-tuned model that has been produced through the initiative can speed up administrative tasks in the healthcare sector. It also promises to ease regulatory compliance in the process. “Cognizant’s offerings include industry-specific solutions like healthcare LLMs and digital twins for manufacturing, while other offerings focus on general AI-driven automation and knowledge management,” Rustagi said. 

To help clients turn LLMs into agents, Cognizant is building what it describes as an agent architecture optimized for modularity. The architecture will include prebuilt components such as integrations with the applications that an agent needs to use while carrying out tasks. Cognizant also plans to add in security guardrails.

“When designing and implementing customized enterprise agentic systems for our clients, we give them a choice over which models to use and adopt the best Gen AI solutions for their business needs,” Rustagi said.

After working with a client to build a set of AI agents, Cognizant can help combine them into so-called multiagent systems. Those are applications that automate multiple tasks by applying a different AI to each chore. Cognizant will provide pre-built multi-agent networks along with a low-code framework for building custom networks. Additionally, it will use Nvidia’s RAPIDS data science libraries to speed up the Apache Spark workflows with which companies supply information for their AI models. Cognizant “helps provide a ‘platform experience’ at reduced production costs,” Rustagi said.

text

text