AI agents that can take entire tasks off users’ hands — booking meetings, resolving support tickets, managing workflows without constant oversight — have become the promise every enterprise vendor is chasing.
But while getting one agent to perform is straightforward enough, getting dozens to coordinate without ballooning costs or security gaps is where most projects stall.
The bottleneck isn’t intelligence. It’s orchestration.
Analysts project the AI orchestration market will surge from $11 billion in 2025 to $30 billion by 2030, but Deloitte predicts that if enterprises orchestrate agents better and address associated challenges, this market could grow by 15% to 30%, reaching as high as $45 billion by 2030.
The orchestra grows louder
The tech giants dominate the market cacophony: Microsoft has deployed its Copilot Studio across more than 230,000 organizations, creating “digital coworkers” that operate without constant human oversight. This month, it’s rolling out agent-to-agent coordination, allowing AI agents to call other agents as tools for more sophisticated workflows.
Salesforce’s Agentforce platform claims to resolve up to 85% of customer service queries autonomously. The platform has emerged as Salesforce’s fastest-growing organic product, reaching a reported 18,500 customers — more than 9,500 on paid plans — with annual recurring revenue (ARR) of $540 million, up 330% year-over-year.
Google’s Vertex AI Agent Builder’s Agent Development Kit has been downloaded more than 7 million times, powering multiagent systems across its products.
OpenAI entered the orchestration race in February with Frontier, a platform for building and governing enterprise AI agents — though some analysts question whether its single-vendor approach will appeal to enterprises increasingly favoring multimodal flexibility.
And now, Lenovo has spoken with TechInformed about the orchestration strategy it unveiled at CES in January 2026, where Chief Technology Officer Tolga Kurtoglu discussed “intelligent model orchestration.”
The gap between ambition and execution
Deloitte’s 2025 Tech Value Survey found that while 80% of organizations believe they have mature capabilities with basic automation, only 28% believe the same for AI agent-related efforts.
More concerning, Gartner predicts more than 40% of agentic AI projects could be canceled by 2027 due to unanticipated cost, complexity or unexpected risks.
Lenovo hopes to close this maturity gap with its Agentic AI platform and three xIQ delivery systems, designed to deploy and manage AI agents across organizations in as little as three months. The company is betting its end-to-end approach, which spans personal devices, data center infrastructure and cloud services, will differentiate it in the crowded market.
For example, the xIQ Agent Platform enables no-code creation of AI agents; the xIQ Digital Workplace Platform improves employee productivity through automated issue detection and resolution; and the xIQ Hybrid Cloud Platform combines operations management, financial optimization and development tools for multicloud environments.
The company worked with dairy producer Yili Group to implement voice-of-customer analytics, supply chain visibility tools and digital commerce systems.
“Yili is accelerating innovation and building a more intelligent supply chain, with Lenovo providing strong support and partnering with us to explore new opportunities for leveraging AI across our operations,” said Tiger Shang, general manager of the data technology center at Yili Group.
“These AI capabilities have improved employee efficiency and helped us better serve our customers.”
Practicing what it preaches
The technology extends to Lenovo’s own operations. The company uses an AI system called iChain to create a digital twin of its global supply chain, which ships more than one device per second to 180 markets worldwide.
Linda Yao, VP and General Manager of Hybrid Cloud & AI Solutions, tells TI that the system has delivered a significant reduction in lead times and a reduction in quality defect costs.
“One of the things I’m most excited about is that AI is moving beyond the phase of fear, hype and complexity,” Yao said. “We’re now seeing real results and real ROI from enterprises that design solutions with a clear end goal in mind.”
The company also unveiled Lenovo and Motorola Qira, a personal AI assistant designed to coordinate tasks across PCs, smartphones, tablets and wearables. Qira won a “Best of CES” award, and Lenovo followed with record quarterly revenues of $22.2 billion in its Q3 results announced in February — up 18% year-over-year.
Kurtoglu outlined the company’s vision for intelligent model orchestration — technology that selects the most appropriate AI models based on user context and intent while optimizing for security, latency and cost.
“When we talk about orchestration, there are actually several different flavors,” Kurtoglu said. “Some of that capability is something we’re building ourselves at Lenovo, but the wider industry is also working on similar concepts.”
The orchestration imperative
Multiple approaches to orchestration are emerging simultaneously, each with individual philosophies.
Microsoft’s Copilot Studio enables “generative orchestration,” where AI dynamically selects tools, topics, agents and knowledge sources based on context. The company has begun experimenting with consumption-based pricing that charges for outcomes rather than seat licenses.
Salesforce’s Agentforce focuses on CRM-centric roles like customer service and sales qualification.
Its Atlas Reasoning Engine delivers enhanced accuracy through refined retrieval-augmented generation.
Google’s Vertex AI Agent Builder offers framework flexibility, allowing developers to build agents with Google’s own Agent Development Kit or deploy existing agents built with LangChain or LangGraph.
The platform’s Agent2Agent protocol, backed by more than 50 partners including Salesforce, ServiceNow and UiPath, enables agents across different frameworks to communicate together.
The cost of getting orchestration wrong is becoming clearer. AT&T, processing 8 billion tokens daily, recently rebuilt its orchestration layer around a multiagent stack where LLM “super agents” direct smaller “worker” agents. The result: 90% cost savings and the capacity to handle 27 billion tokens daily — more than a threefold increase.
Research suggests that today’s multiagent systems perform better with humans in the loop, benefiting from human experience while remaining aligned with organizational expectations.
Deloitte predicts a progressive “autonomy spectrum” will emerge, ranging from humans-in-the-loop to humans-on-the-loop to humans-out-of-the-loop — based on task complexity, business domain and outcome criticality.
At Lenovo, Kurtoglu acknowledged that the technology remains at an early stage. Model orchestration, task orchestration and agent orchestration are all developing at different speeds, he said, and the industry has yet to mature in enabling seamless information exchange between AI agents.
Flexibility and deployment
According to Yao, the company serves eight of the world’s 10 largest cloud providers while also supporting on-premises and private cloud deployments for enterprises with strict data sovereignty and privacy requirements.
“Enterprise IT environments are not one-size-fits-all, so flexibility is essential,” Yao said. “AI has added a new dimension to this. In many ways, it’s been a catalyst, prompting customers to prioritize investment in their hybrid environments.”
The healthcare sector illustrates the range of deployment models, she added. In Mexico City, the company helped Hospital Angeles build its own LLM for patient diagnostics while keeping all data on-site.
“Healthcare organizations are extremely sensitive to patient data privacy and regulatory compliance,” Yao added. “They need access to high-performance compute and advanced technologies, but in a way that remains fully on-premises and under their control.”
From productivity to revenue
Looking ahead, Yao believes the next frontier extends beyond productivity gains to revenue growth.
Lenovo has deployed AI agents across its contact centers, and it claims to see double-digit productivity improvements alongside double-digit revenue uplift and higher customer satisfaction scores.
Gartner predicts that by 2028, 33% of enterprise software applications will include agentic AI, with at least 15% of day-to-day work decisions made autonomously through AI agents. Meanwhile, a global survey of 200 HR leaders found that 86% of chief human resources officers see integrating digital labor as central to their role. And as employers start to iron out roles for each AI agent, 44% of leaders now expect AI agents to take lead roles in managing specific projects with human team members within the next two to three years, according to KPMG.