Sequoia Capital: This is AGI!

On January 14, Sequoia Capital partners Pat Grady and Sonya Huang published an article titled “2026: This is AGI,” stating that Artificial General Intelligence (AGI) is no longer a distant future but has become a reality with the emergence of long-horizon agents. Although there are still disagreements regarding the technical definition of AGI, the article emphasizes that AI capable of autonomously solving problems is now operational, and 2026 will be its year.

Sequoia Capital identifies coding agents as the first instance of AGI, with more types of agents emerging. Unlike early conversational AIs, the new generation of long-horizon agents can reason based on baseline knowledge and achieve goals through continuous self-iteration. This shift marks a transition of AI from mere “conversationalists” to “executors” capable of delivering actual work.

This transformation will have profound implications for business and investment. Sequoia Capital believes that as agent capabilities grow exponentially, the logic of founders building products will fundamentally change—from selling software to directly “selling work outcomes.” Future AI applications will no longer just be auxiliary tools but entities that can work alongside humans as “colleagues,” with users evolving from independent contributors to managers of agent teams.

With Claude Code and other coding agents recently surpassing critical capability thresholds, market perceptions of AGI have been reshaped. The article highlights that through reinforcement learning and optimized agent architectures, agents’ abilities to handle complex tasks are doubling every seven months, fundamentally altering corporate talent structures and productivity boundaries.

Functional Definition: AGI as the Ability to Solve Problems Autonomously

Sequoia Capital states that as investors, they do not intend to engage in the technical definition debate of AGI but propose a pragmatic functional definition: AGI is the ability to solve problems autonomously. For businesses aiming to succeed, how AI achieves its goals is less important than whether it can genuinely complete tasks.

The article breaks down AI with this capability into three core elements:

• Baseline Knowledge (Pre-training): This was the driving force behind the 2022 ChatGPT moment.
• Reasoning Ability (Inference Calculation): Achieved with the release of the o1 model by the end of 2024.
• Iterative Ability (Long-horizon Agents): This latest breakthrough allows AI to work autonomously like general intelligent humans, correcting errors and deciding on next steps without specific instructions within hours.

From Instructions to Autonomy: The Work Loop of Agents

To illustrate what it means to “solve problems autonomously,” the article uses a hiring scenario: When a founder needs to find a developer relations head who understands technology and is active on social media, the traditional approach is to post a job description. In contrast, agents can autonomously execute complex search loops.

According to the article, agents can complete the psychological loop of a human hiring expert in 31 minutes: They will not only search for relevant positions at competitor companies like Datadog and Temporal on LinkedIn but will also turn to YouTube to filter high-engagement speakers and further cross-reference activity and content quality on Twitter. Agents can even detect potential resignation signals by analyzing a decline in posting frequency, ultimately identifying the best candidate and drafting a personalized outreach email.

This ability to establish hypotheses, test, iterate, and adjust direction until achieving goals in ambiguous environments is the core feature of long-horizon agents. Although they still produce hallucinations or lose direction at times, their developmental trajectory is irreversible, and errors are becoming increasingly correctable.

Technical Pathways: Dual Drivers of Reinforcement Learning and Agent Architectures

Regarding how this leap is achieved, Sequoia Capital notes that enabling models to think for extended periods is not easy. Currently, two technical pathways have proven effective and scalable:

One is Reinforcement Learning, primarily led by research labs. Through continuous “nudging” and guidance during training, models are taught to maintain focus over long periods. Significant progress has been made in multi-agent systems and the reliability of tool usage.

The second is Agent Architectures, which belong to the application layer. Developers design specific scaffolds (such as memory handover, compression, etc.) to circumvent known limitations of models. Currently well-regarded products in the market, such as Manus, Claude Code, and Factory’s Droids, benefit from their excellent architectural design.

According to METR’s tracking of AI’s ability to complete long-horizon tasks, progress in this field is growing exponentially. Based on current trends, agents will reliably complete tasks that human experts currently take a full day to accomplish by 2028, and by 2034, they will be able to handle a year’s worth of work.

Business Transformation: From Software to “Digital Employees”

“Can you hire an agent?” Sequoia Capital sees this as the litmus test for AGI. The current market landscape indicates that specialized agents are rapidly emerging across various sectors, from pharmaceuticals with OpenEvidence, legal with Harvey, to cybersecurity with XBOW.

This signifies a massive paradigm shift for entrepreneurs. AI applications in 2023 and 2024 are largely “conversationalists” with limited impact; however, applications from 2026 onwards will be “executors.” This transition makes “selling work” a possibility. Founders need to rethink which ongoing tasks can be taken over by agents and how to price and package based on “results” rather than “tools.”

The article concludes with a call for the market to “saddle up” for the exponential growth of long-horizon agents. While today’s agents may only reliably work for about 30 minutes, they will soon be able to handle a full day’s workload and eventually tackle tasks equivalent to a century’s worth of human work.

This means that what was once considered an overly ambitious roadmap—such as cross-referencing 200,000 clinical trial data or reconstructing the entire U.S. tax code—has now become feasible. In the year of AGI, ambitious plans are gradually transforming into realistic business strategies.