Sequoia Capital: 2026 will be the inaugural year of AGI, and programming agents have fired the first shot!
```
Artificial General Intelligence (AGI) is no longer a distant future; it has become a reality with the emergence of "long-horizon agents." According to a June 14 article by Sequoia Capital partners Pat Grady and Sonya Huang titled "2026: This is AGI," although there is still technical debate over the definition of AGI, functionally, AI with autonomous problem-solving capabilities has officially arrived, and 2026 will belong to them.
Sequoia Capital states that coding agents are the first practical instance of AGI, and more agent types are rapidly emerging. Unlike early conversational AI, the new generation of long-horizon agents can, like humans, reason based on baseline knowledge and achieve goals through continuous self-iteration. This leap marks the transition of AI from mere "conversationalists" to actual "executors" who can deliver work.
This transformation will have far-reaching impact on business and investment. Sequoia's analysis suggests that as agent capabilities grow exponentially, the logic by which founders build products will fundamentally change—from selling software to directly "selling work outputs." Future AI applications will no longer be just auxiliary tools, but rather entities capable of working as "colleagues" around the clock, with users shifting from independent contributors to managers of agent teams.
With Claude Code and other coding agents recently crossing key capability thresholds, market perception of AGI has been reshaped. The article stresses that through reinforcement learning and optimization of agent architectures, agents' ability to handle complex tasks is doubling every 7 months, which will completely redefine enterprise talent structure and productivity boundaries.
Functional Definition: AGI as the Ability to "Solve Problems Autonomously"
Sequoia Capital states that as investors, they have no intention of engaging in technical disputes over the definition of AGI, instead proposing a pragmatic functional definition: AGI is the ability to "solve problems autonomously". For ambitious businesses, how AI achieves its goals is irrelevant; what matters is whether it can actually get the job done.
The article breaks down the core elements of such AI into three components:
Baseline knowledge (pre-training): This was the core driving force behind the "ChatGPT moment" in 2022.Reasoning ability (computation during inference): Achieved with the release of the o1 model at the end of 2024.Iteration ability (long-horizon agent): The latest breakthrough: AI can now work autonomously for hours like generally intelligent humans, correct errors, and decide next actions without explicit instructions.
From Instructions to Autonomy: The Closed-Loop Workflow of Agents
To illustrate "autonomous problem-solving," the article uses a recruitment example: When a founder wants to hire a developer relations lead who is both technical and active on social media, the traditional approach is to post a job description. By contrast, an agent can autonomously execute a complex search loop.
According to the article, an agent can complete a human recruitment expert’s mental cycle in just 31 minutes: Not only does it search competitor companies like Datadog and Temporal on LinkedIn, but it also goes to YouTube to filter for speakers with high engagement, and further cross-references Twitter for activity and content quality. The agent can even detect potential resignation signals by analyzing a drop in posting frequency, ultimately selecting top candidates and drafting personalized outreach emails.
This ability to hypothesize, test, experiment and adjust direction in a fuzzy environment until a goal is achieved is the core trait of long-horizon agents. Though agents may still hallucinate or get lost, their trajectory is irreversible, and errors are increasingly correctable.
Technical Pathways: Dual Drive of Reinforcement Learning and Agent Architectures
On how to achieve this leap, Sequoia Capital points out that getting models to think for extended periods is no easy feat. Two technical paths have proven effective and scalable so far:
First is Reinforcement Learning, mainly driven by research labs. Through constant "nudging" and guidance, models are taught to stay focused over long periods during training. Significant progress has been made in multi-agent systems and reliable tool use.
Second is Agent Harnesses, in the application layer. Developers design specific scaffolding (like memory handoff, compression, etc.) to circumvent known model limitations. Products praised on the market, such as Manus, Claude Code, and Factory's Droids, all owe much to outstanding architectural design.
According to METR’s tracking of AI long-horizon task completion capabilities, growth in this field is exponential. Projecting current trends, by 2028 agents will reliably complete a full day’s worth of expert human work, and by 2034 will be able to accomplish a year’s workload.
Business Transformation: From Software to "Digital Employees"
"Can you hire an agent?" Sequoia Capital believes this is the litmus test for AGI. The current market shows that specialized agents are quickly appearing across all industries—from OpenEvidence in pharma, Harvey in law, to XBOW in cybersecurity.
This means a tremendous paradigm shift for entrepreneurs. AI applications in 2023 and 2024 were largely "conversationalists" with limited impact; but from 2026 onward, they will be "executors". This shift makes it possible to "sell work." Founders need to rethink: In this new paradigm, what tasks requiring ongoing attention can be turned over to agents? How can pricing and packaging shift focus from "tools" to "results"?
The article ends by urging the market to "Saddle Up" and prepare for exponential growth in long-horizon agents. Though today’s agents are reliably working for about 30 minutes, soon they’ll take on a full day’s workload, and ultimately, tasks equal to a human century of work.
This means what was once seen as overly ambitious—like cross-referencing 200,000 clinical trial data sets or restructuring the entire US tax code—has now become attainable. In the year of AGI, ambitious planning is rapidly turning into practical business strategy.
Risk Warning and DisclaimerThe market has risks; investment needs caution. This article does not constitute personal investment advice nor consider the specific investment objectives, financial situation, or needs of any individual user. Users should consider whether any opinions, views, or conclusions herein fit their particular circumstances. Investment is at your own risk.
```