AI-first seems to be the way forward for most tech companies. In late May, Lenovo, most well known for its devices such as laptops, announced that revenue for 2025/2026 reached a total of $83.1 billion, a part of which can attributed to the company's prolonged investment in what it has described as hybrid AI.
The company noted that AI-related revenue grew 84% year-over-year under the company's strategy, which has been deployed to engage the consumer and enterprise markets.
The company maintains its device business, offering personal devices across laptops, tablets, smartphones, and more. It also has an infrastructure business, which provides companies with access to data centers, high performance computing, servers, networking, storage, plus the software stack that sits on top. And now also has a fast-growing services business that provides fully-integrated AI solutions.
Going AI-First
Implementing AI across companies can prove tricky in some aspects. For instance, Microsoft is reportedly moving away from Claude Code after exhausting its annual token-based billing budget. Uber's COO also noted that the company exhausted its entire 2026 AI budget in just four months.
If AI adoption isn't carefully monitored, the costs can quickly spiral. In this context, International Business Times spoke with Tolga Kurtoglu, chief technology officer at Lenovo, to discuss its AI initiatives.
Kurtoglu noted that Lenovo has undergone a complete transformation from a device maker into an AI-first company. "We are known as a device company but we're going through a fundamental shift, I would say, as a company, to moving from being a hardware or device first company to being an AI first company," he said.
"Our vision of AI is grounded in what we call hybrid AI, and the simplest way to describe that is, whether you're in personal AI or in enterprise AI, you can think of the world as a distributed set of devices, a distributed set of data, that sits in different locations, a distributed ecosystem of models, big and small, and agents, and compute that needs to happen," Kurtoglu added.
Hybrid AI is about orchestrating all of those angles, bringing AI experiences to consumers and enterprises alike. More specifically, building agentic compute and AI experiences that simplify the intermediation and orchestration of data, devices, and compute, plus the models they interact with in the environment.
Pivoting Toward AI-Driven Operations
Investing in AI agents is a critical step on the road toward becoming AI-first. However, it is easier said than done.
A Sinch survey of more than 2,500 senior decision-makers found that three-quarters of enterprises have rolled back or shut down a customer-facing AI agent after deployment. The majority of rollbacks were due to governance failures. Another survey from Solvd found that while 90% of CIOs and CTOs expect experimental AI investment to increase, 72% believe AI projects will be shut down in the next year for failing to meet KPIs.
Lenovo has addressed the adoption of AI by building a new internal platform group known as Lenovo AI Technology Center (LATC). LATC builds some of the company's core technology capabilities, including model deployment, orchestration, security, and privacy.
The company pivoted toward identifying the foundational core technologies needed to help the platform group innovate faster. Kurtoglu notes that "tokenmaxxing," a metric used by some companies to monitor developer productivity related to AI, hasn't been part of the company's core focus.
"Just because you're maxing out on the consumption of your tokens does not necessarily mean that translates into business outcomes and impact."
Instead, the company has focused on putting in place the processes needed to fuel the adoption of AI coding tools, and ensuring the compute consumed by adopting AI tools translates directly into business impact. He says that "we're constantly pushing all the teams to essentially adopt AI in our internal workflows."
In terms of tools being used by Lenovo's developers, Kurtoglu pointed to Cursor and Microsoft GitHub Copilot as go to choices, but said he wasn't sure to the extent that Claude Cowork or Claude Code has been used internally. The company has also launched a rapid prototyping group to experiment with emerging technologies.
Introducing The Super Agent
In the realm of digital transformation, technology adoption often becomes a go big or go home effort. Lenovo isn't just aiming to roll out AI agents in the enterprise, but to go a step further and deploy what Kurtoglu described as super agents.
Kurtoglu says super agents can be thought of as the entry point of the agent's user interface and the entry point of being able to perform orchestration. In this sense, a super agent has dynamic model orchestration capabilities.
So any time a user interacts with a task, the company wants to enable the user to connect to the best model, whether it's local or in the cloud, depending on the user and intent. Lenovo is building a model orchestration capability that enables it to do just that, deployed with Lenovo Qira and Motorola Qira.
Lenovo Qira and Motorola Qira sit on the personal AI side of the product portfolio, what Kurtoglu describes as a cross-drive ambient intelligence agent, a layer that sits in the background and simplifies the day-to-day tasks of the end user. It's defined as ambient because it engages only when the user wants.
"Think of it as almost like a conductor in an orchestra, so it knows which agent to go to, which database to query from, or which data sources to combine, which models to send the prompts to, etc., so it acts, but by no means by itself, by knowing everything, but really leveraging all the capabilities that it is connected to, and that's the definition of a super agent," Kurtoglu said.
The super agent thus has connections to connectors, across MCP servers, first party applications, third party applications, local models that have been deployed on the device or across different devices, integrating across memories, the personal knowledge base, context windows to provide one unified personal foundation.
A super agent thus isn't just a standalone agent, but something that offers additional orchestration capabilities, the ability to trigger other agents. Agent collaboration and multi-agent orchestration are key pillars to the super agent.
Kurtoglu shares that Lenovo also provides a super agent tool internally to all of its 70,000 employees – a single gateway that provides access to multiple models and multiple categories of enterprise data, legal, HR policies, finance, etc. This agent is designed to act as an orchestrator or connector.
Reaping The Rewards Of AI
While Lenovo initially deployed this super agent as an employee assistive tool, its orchestration capabilities are also becoming part of the company's external enterprise AI offerings. AI has also proved itself to be impactful across a range of internal operations such as supply chain and operations.
For example, Kurtoglu noted , Lenovo does over 90% of its manufacturing through its manufacturing facilities, factories, and supply chain. The company has thus deployed agentic capabilities to assist in planning, forecasting, and execution of manufacturing operations, as well as supply chain operations.
In this context, stakeholders can measure productivity gains, yield, time and speed, flexibility and robustness in supply chain availability, and other KPIs. He noted the company has seen significant improvements and value from iChain — its supply chain super agent.
"I think overall we're very excited about the hybrid AI vision, and where we are with our journey," Kurtoglu said. "I am particularly excited, as the CTO, that we see the initial commercial launches that comes from that we were first when it comes to AI PC, which was the most basic deployment of AI on edge devices like personal computers, but now we have evolved from just deploying AI on a device to building ambient intelligence that can operate across devices."
The move highlights that agentic AI isn't just impactful in the enterprise but the consumer market too. While there are risks to be ironed out, endeavoring to define governance from the get go is one way to increase the potential value obtained from adopting the technology.
