AI agents are quickly becoming a standard in various industries. According to a recent LangChain report, while 51% of companies are already using AI agents in production, 78% have plans to adopt them in the near future.
Originally rooted in robotic process automation (RPA), which handles simple, repetitive tasks, AI in the supply chain is now evolving rapidly.
Today’s AI agents go beyond task automation—they’re capable of managing entire workflows, adapting in real time and supporting decision-making across logistics, procurement, and inventory.
Manhattan’s Alternative Approach
While many vendors are racing to build AI agents for their clients, Manhattan Associates—a global technology leader in supply chain and omnichannel commerce—has chosen a different route.
Manhattan Associates India Development Centre is the largest GCC with over 2,000 employees, having started with just five employees in 2002.
Rather than pre-building large sets of agents, the company has introduced Manhattan Agent Foundry—a platform that enables clients to create their own agents, tailored to specific needs and workflows.
“You can write your agent, you can take our agents, you can mix and match, you can work with it,” Ushasri Tirumala, executive vice president and GM, India at Manhattan Associates, said in a recent interview with AIM.
“That is the philosophy that we have. We are not announcing we will do 200 agents or whatever. We are going to provide the critical agents for you to understand and leverage them, and then you can develop your own.”
What Makes Manhattan Agent Foundry Stand Out
Agents built on the Manhattan platform are powered by large language models (LLMs) and integrated into a cloud-native, microservices-based API architecture. These AI agents are not just reactive—they can orchestrate processes, adjust dynamically to real-world conditions, and communicate across systems.
Notably, Manhattan’s agents are designed to interoperate with third-party agents, adhering to standards like A2A and MCP. This means enterprises can mix Manhattan-built agents with those developed or acquired elsewhere, creating a truly interconnected, AI-driven ecosystem.
“Customers would have the ability to get the information when they want, however they want, and to make decisions,” Tirumala shared.
“It is very important for companies because it would help in certain areas so that you don’t need to do the kind of context jobs…using more skilled manpower that would go away,” she added.
Tirumala confirmed that these agents are being integrated across all Manhattan product lines, with broader availability planned by this fall.
A Unified View of the Supply Chain
Manhattan’s product philosophy revolves around unification. From demand forecasting to order and transportation management, the focus is on creating a cohesive view across supply chain operations.
“More and more today, we are talking about unification. It is not enough if you look at one transportation management or distribution management,” Tirumala explained. “You have to have a total view, and [know] how to bring that visibility to people who are planning, using, etc. That is what we are bringing.”
Manhattan’s broader product strategy already includes several key innovations aimed at improving supply chain intelligence and customer experience.
These include Manhattan Active Supply Chain Planning (MASCP), designed for advanced forecasting and supply optimisation, and the Enterprise Platform Framework (EPF), which enables seamless omnichannel visibility.
Manhattan Active Maven, a generative AI solution tailored for next-generation customer service, and Manhattan Assist, an AI-powered assistant that provides contextual guidance on product functionality, API structures, and more.
Furthermore, Manhattan’s move into AI didn’t happen overnight. The company began investing in AI years ago and even developed its own machine learning platform.
“Even a couple of years back, we had developed our own machine learning platform,” Tirumala noted.
Talking about LLMs, Tirumala revealed that Manhattan also maintains a strong partnership with Google, hosting its solutions on Google Cloud and leveraging Google’s AI infrastructure.
“We do not stick to one LLM,” Tirumala added. “We work with Google in a big way. We work with Google platforms. So, our solutions are also hosted on the Google Cloud.”
Bengaluru for Scaling
Manhattan Associates takes a deliberate and focused approach to team structure. Unlike companies with multiple development centres across the globe, Manhattan operates from just two primary locations—the United States and India—with over two-thirds of its R&D workforce based in Bengaluru.
“We are not in the tens of thousands in number. And we believe that product development requires very close working together,” Tirumala said “We cannot have one centre here, another centre somewhere else…That will be quite difficult.”
Tirumala mentioned that an earlier internal study even explored adding a third development hub, but the findings supported sticking with Bengaluru as the optimal location.
“We looked at whether it is worthwhile establishing another centre from a talent perspective, not for anything else. And then that study came back with the finding that being in Bengaluru is the best thing. Not even Mysuru at that time,” she said.
With a lean team of just over 2,000 employees, Manhattan India handles not only R&D but also customer support and product-based services—from implementation to lifecycle support.
“Product-based services, the implementation of that, the extension of that, and improving whatever the customer wants from their work perspective—all of those things are carried out here,” Tirumala noted.
The post Why Manhattan Associates is Not Developing AI Agents for Its Clients appeared first on Analytics India Magazine.
