In a recent presentation at Y Combinator’s AI Startup School, Andrej Karpathy made a bold yet grounded observation:
‍

“2025 is not the year of agents. But 2025–2035 will be the decade of agents.”

‍

While the excitement around AI agents is currently peaking—especially in Silicon Valley—Karpathy urges us to take the long view. The most meaningful breakthroughs won’t happen this year. They’ll unfold over the next decade.

‍

Why Not 2025?

Despite impressive demos and beta tools, today’s AI agents are still limited. Tools like OpenAI’s Operator can assist with simple tasks (scheduling, drafting, querying), but they fall short in:

• Managing long-horizon tasks
• Performing reliably in real-world settings
• Handling complex, multimodal information

The leap from utility to autonomy—from assistants to full-scale operators—will take years of advancement in infrastructure, reliability, and AI-human collaboration.

‍

What the Decade Will Bring (2025–2035)

Karpathy paints a compelling picture of what’s coming:
A world where organizations deploy fleets of AI agents to manage complex, long-running operations. Humans won’t be replaced but elevated—acting as high-level supervisors to AI systems that execute and optimize work in real time.

‍

“Tomorrow, you’ll spin up organizations of Operators for long-running tasks of your choice (e.g., running a whole company). You could be a kind of CEO monitoring 10 of them at once…”
— Andrej Karpathy

‍

What Needs to Evolve

To make this vision reality, the next decade must focus on:

1. Multimodal Capabilities
   Agents will need to reason across text, voice, images, and video to better understand their environment and tasks.
2. Long-Term Planning
   Agents must handle multi-step workflows over hours, days, or even months—beyond today’s stateless commands.
3. Collaborative AI Systems
   Multi-agent orchestration will allow groups of agents to divide and conquer problems, handing off tasks intelligently.
4. Human-in-the-Loop Governance
   While agents gain autonomy, oversight mechanisms will be crucial for ethics, alignment, and accountability.

‍

Implications for Business & Society

From customer service and operations to healthcare and education, agentic AI will transform how work gets done.
As the market for AI agents surges from $5.3 billion in 2023 to an estimated $200+ billion by 2035, the potential is massive—but so are the responsibilities:

• Regulatory oversight
• Workforce transition
• AI safety and transparency

‍

Final Takeaway

2025 may not be the year of agents.
But it is the start of a decade that will reshape work, leadership, and the human-machine dynamic.
The agentic future isn’t a hype cycle—it’s a horizon.

Is your organization building for the decade of agents?

‍

‍

FAQs

‍

1. Why does Andrej Karpathy say 2025 is not the year of agents?
While we’ve seen early progress, today’s AI agents are still limited in handling long, complex tasks. Multimodal integration, long-term planning, and reliability still need to mature before agents can scale across industries.

‍

2. What’s the significance of the 2025–2035 decade for AI agents?
This is the horizon in which agents evolve from simple task executors to fully autonomous systems that can run departments, manage workflows, and collaborate with minimal human oversight.

‍

3. What advancements are needed for agentic AI to scale?
Key areas include:

• Seamless multimodal processing (text, audio, video, images)
• Long-horizon planning and memory
• Multi-agent collaboration frameworks
• Strong human-in-the-loop governance mechanisms

‍

4. What does “running a whole company with agents” mean?
Karpathy envisions a model where one human supervises several autonomous agents, each handling business functions like operations, finance, or customer service—similar to being a CEO managing autonomous departments.

‍

5. What’s the projected market size for AI agents?
The AI agent market is expected to grow from $5.3B in 2023 to over $200B by 2035, driven by adoption across sectors like healthcare, finance, education, and enterprise automation.

‍

6. Will humans become obsolete in an agent-driven world?
No. Human roles will shift from execution to orchestration—supervising, resolving edge cases, and setting strategic direction while agents handle operational tasks.

‍

7. What are the societal risks and responsibilities in this shift?
Key challenges include:

• Ethical governance of autonomous agents
• Workforce reskilling and transition
• Ensuring transparency, fairness, and safety in AI decision-making

‍

#AI #Agents #AndrejKarpathy #FutureOfWork #Automation #DecadeOfAgents

‍

‍