External signal·Sequoia Capital·Jan 14, 2026·Pat Grady, Sonya Huang·12 min read
2026: This is AGI
Summary
Sequoia partners Pat Grady and Sonya Huang argue that long-horizon agents—AI systems that can plan, use tools, keep memory, take actions, evaluate outcomes, and loop until a goal is met—are functionally AGI, and that 2026 is their year. They adopt a pragmatic definition of AGI as the ability to figure things out and solve problems autonomously, tracing three stages: the 2022 ChatGPT foundation of broad language competence, the late-2024 inference-time-compute breakthrough with OpenAI's o1, and the late-2025 arrival of capable agents such as Claude Code. They predict AI products will shift from chatbots used a few times a day to all-day 'doers' that feel like colleagues. Their litmus test, attributed to Sarah Guo, is whether you can hire an agent.
Predictions for the future of work
Grady and Huang predict that in 2026 and 2027 AI agents will function less like tools and more like colleagues you delegate to, with interfaces moving from chat to 'hiring an agent' and multiple agent instances running in parallel all day. They foresee agents taking on multi-step knowledge work such as recruiting, research, and coding end to end, implying substitution for tasks currently done by white-collar workers and a large commercial opportunity. The horizon is explicitly near-term, framed as already underway; the affected parties are knowledge workers and the firms that deploy agentic labor.
Originally published by Sequoia Capital · Jan 14, 2026
Read the original at Sequoia Capital