Systems integrators and services companies are launching AI agents, releasing frameworks and trying to help enterprises build multi-agent systems. The big question is whether AI agents turn out to be a boon or a bust for systems integrators in the long run.
In recent days, we've heard from multiple systems integrators with more on tap talk about agentic AI. The extension of integrators into agentic AI makes sense given that they have the expertise to work across systems and processes. Consider:
Kyndryl, a services provider focused more on infrastructure, released the Kyndryl Agentic AI Framework, which orchestrates and dispatches AI agents that respond to shifting conditions. The framework is a way for Kyndryl to move up the stack to higher level offerings because it moves the integrator beyond infrastructure to workflows and processes.
According to Kyndryl, its Agentic AI Framework leverages algorithms, self-learning, optimizations and AI agents to run applications and processes.
Wipro said on its first quarter earnings call that enterprises are shifting discretionary funds to data and AI modernization. "AI is no longer a niche. It's becoming essential to how businesses operate at scale," said Wipro CEO Srinivas Pallia.
He added:
"Our AI capabilities are integrated into both industry and cross-industry solutions. By combining domain expertise with AI, we are able to deliver value through solutions such as hyper-personalized wealth management and predictive industrial insights. We have deployed over 200 AI-powered agents using advanced technologies from leading hyperscalers. These agents enables smarter lending, intelligent claims processing and autonomous network management."
At AWS Summit New York, there were multiple partners talking about the foundation needed for AI and agentic AI adoption. Deloitte's Chris Jangareddy, Managing Director of the company's AI, GenAI and Data Engineering, said the company will have nearly 180 agents on AWS Agent Marketplace.
- AWS' practical AI agent pitch to CxOs: Fundamentals, ROI matter
- AWS launches Bedrock Agent Core, custom Nova models
- Intuit starts to scale AI agents via AWS
- AWS launches Kiro, an IDE powered by AI agents
According to Jangareddy, these agents are aimed at business problems, processes and specific tasks. Deloitte's AI agents are designed to be reusable Lego locks that will ultimately make up multi-agent systems. One offering is AI Advantage for CFOs that serve as a digital twin for CFOs, he said. The agents are built on Deloitte's institutional knowledge base of queries that are now prompts.
"These are not licensed, but are for clients," said Jangareddy, who noted that Deloitte is looking to transform its model from traditional billing to an outcome-based approach.
In a demo, Deloitte outlined Zora AI, which is part of an effort to produce AI agents that are product offerings. Deloitte views AI agents as digital labor that focuses on executing on processes. Zora AI is also integrated with SAP Joule.
AWS’ Brian Bohan, Director, Global Lead, Consulting Partner Center of Excellence, said during a talk that companies automating multiple business processes with agentic AI are seeing 30% to 40% productivity gains. He expects more efficiency to be unlocked.
Why? The cost of models is falling as are training and inference expenses. However, many AI projects aren't scaling due to a lack of architecture, data infrastructure and expertise. "There's just the complexity of integration," said Bohan.
Bohan added that change management, workflow optimization and the pace of innovation are all challenges. Enterprises will get to multi-agent systems across functions like finance, procurement and supply chain.
It's clear that systems integrators see AI agents as a booming business as well as a way to transform their businesses. The flip side of this transformation is that AI agents may ultimately hamper the systems integrator model.
Constellation Research analyst Holger Mueller said:
"As with any new technology, enterprises are looking at system integrators for help adopting them and AI is no difference here. The question is whether AI is so strategic that enterprises need AI skills inhouse, or can they rely on the integrator model. The experience depth is low for anyone as no one has more than two years in genAI experiences. Or more than 10 projects. It is likely going to be strategic for enterprises to have their own AI capacity and competency, especially once we move to inter-enterprise agents and the uptime and capability of frontline and backend agents determine success."