Theme

Agents counted as headcount

Explore this theme’s evidence over time →

Sources

Attribution and reporting linked by this record; links do not imply independent verification.

Welcome to January 13, 2026 (theinnermostloop.substack.com)

The shift from agents as tools a firm uses to agents as staff a firm employs — counted, budgeted, targeted for a ratio against humans, and eventually managed by the same org chart.

Development timeline

Newest reporting dates first. Reporting dates may differ from event dates.

2026-09-17: Claude brings chat and Cowork together

2026-09-17: An AI store manager dismisses a human worker

2026-08-31: A researcher proposes an org chart where the agent brings humans the decisions

2026-08-27: A software giant puts its whole CRM inside a model, billed by consumption

2026-08-21: A social network burns trillions of tokens a week writing its own software

2026-07-04: The argument shifts from genies summoned per task to entities that persist

2026-07-01: A cloud raises a $1B army of forward-deployed engineers

2026-06-29: Engineering, product and design melt into five archetypes

2026-06-24: A model becomes a persistent asynchronous colleague

2026-06-10: A services giant slows hiring expecting as many agents as humans

2026-05-16: A foreign minister runs his parliamentary affairs through a home-built agent

2026-05-06: An agent becomes the world's first AI cafe owner

2026-05-04: A government directs half its operations onto agents within two years

2026-04-29: An orchestrator gives every open ticket its own running agent

2026-04-23: Background agents build memories from screen captures

2026-04-17: A coding tool is promoted from autocomplete to coworker

2026-04-16: A company exhausts its entire year's AI budget in months

2026-04-13: Meta trains a photorealistic avatar of its founder for employees

2026-04-13: An AI is handed a three-year lease and hires its own staff

2026-04-12: The Linux kernel issues coding assistants a dress code

2026-04-08: Sandbox containment is promoted from research topic to product feature

2026-03-28: At one lab nobody has hand-written code in months

2026-03-23: A chief executive builds an agent to help him be chief executive

2026-03-10: A team of agents reviews every pull request at $15 to $25 a time

2026-02-27: An agent gets a work calendar before most interns do

2026-02-27: A sovereign wealth fund screens investments for ethics with a model

2026-02-26: A founder gives an AI his inbox for fourteen days to raise its own money

2026-02-26: Employees rehearse with an AI clone of their CEO

2026-02-26: An engineer leaves for the weekend and comes back to a shipped feature

2026-02-23: Agents manage one in six US apartments

2026-02-08: Agents get decorative housing and cron jobs to show up for work

2026-02-07: One tool is projected at a fifth of public commits by year end

2026-02-07: Racks of Mac Minis are run as round-the-clock employees

2026-02-07: OpenAI bans its own employees from using editors and terminals

2026-02-06: A model finds 500 zero-days, some decades old

2026-02-06: One tool reaches 4% of all public GitHub commits

2026-01-31: OpenAI uses a model on 600 petabytes to make its own decisions

2026-01-27: Users give their agents Apple IDs and phone numbers

2026-01-26: A CEO asks the model to manage him

2026-01-25: Claude Code gets a project management cortex

2026-01-24: Coding agents scale best in a corporate hierarchy

2026-01-15: McKinsey tests candidates on filtering model output

2026-01-14: An AI team builds and scales real businesses

2026-01-13: McKinsey employs 20,000 agents alongside 40,000 people

Knowledge graphopen in browser →

Cite or correct this record

Citation formatCitation text

Agents counted as headcount
Innermost Loop KB — secondary record
Reporting date: unknown
KB permalink: https://nicholsn.github.io/innermost-loop-kb/themes/agents-on-the-org-chart
Attributed sources:
- Welcome to January 13, 2026 — https://theinnermostloop.substack.com/p/welcome-to-january-13-2026
This citation does not imply independent verification of linked-source claims.

Suggest a correction — opens a GitHub draft for you to review and submit.