Sequoia Says AGI Is Here: How a Century‑Long Vision Could Be Realized in 100 Days

Computing Revolution

Silicon transistors created networks, the Internet solved information distribution, and mobile devices placed the network in pockets. Decades of accumulated compute, bandwidth, and data have moved humanity from an information‑distribution era to an information‑processing era.

Cloud computing expanded the global software market to $650 billion . AI now targets the U.S. legal‑services market of $4 trillion , part of a $10 trillion global services opportunity. The speed of AI‑driven market entry is unprecedented, with many companies reaching $1 billion in revenue within months.

Intelligent Agents

Early 2022 agent projects that used self‑looping settings went viral, often crashing but attracting developers. Today agents are deployed in everyday tasks, such as automated tax‑evasion reporting, and can be assembled by anyone with a smartphone.

Agents consist of three core components:

Brain : reasoning and planning; current models can stay on task for hours without deviating.

Limbs : execute actions by invoking terminals, calendars, search engines, or third‑party apps, dramatically increasing the value of existing software toolchains.

Harness : a reinforcement‑learning‑based control system that provides persistence, allowing agents to learn from failures and continue progressing.

The next wave includes backend asynchronous agents, massive networks of sub‑agents, and “dark‑factory” modes that run without human oversight in security and code‑review pipelines.

Agents offer token‑based billing (low cost), unlimited concurrency (scalability), and continuous operation (endurance), challenging traditional employment models.

Cognitive Revolution

Work is divided into physical (mechanical) and cognitive (mental) categories. Physical work has been mechanized by steam engines, internal‑combustion engines, and electric motors; today over 99 % of physical labor is performed by machines. The authors extrapolate that 99.9 % of cognitive work will soon be handled by neural networks .

Historical analogies illustrate devaluation of intellectual labor: aluminum was once a precious metal, then became cheap packaging after electrolysis, mirroring how AI reduces the cost of expertise.

Examples of domain‑specific agents:

Medical agents that deeply analyze genomes and recommend clinical trials.

Legal agents that draft contracts and can even appear in court.

Mathematics agents that search for novel superconducting materials.

Personal‑assistant agents that fully manage email, calendars, and financial statements.

Productivity case studies cited include:

A top developer completing a three‑year, moon‑level project in a single holiday.

One weekend rewrite of a large customer‑service system.

Six‑week refactor of 8 million lines of code.

Agents can execute complex, long‑duration tasks autonomously: “If you can dispatch an agent to do a job and it can recover from failure and persist until the job is done… that feels pretty much like AGI.”

Quote: “Whatever you can imagine building over the next hundred years, we think is now possible in 100 days thanks to agents.”

Reference: https://www.youtube.com/watch?v=LRo33rnv6rQ