Dan Shipper is the co-founder and CEO of Every, a media and software company exploring work with AI. He writes Chain of Thought, co-hosts The Every Podcast and previously co-founded Firefly, which was sold to Pegasystems in 2014.[1][2]
Start with The Knowledge Economy Is Over. Welcome to the Allocation Economy. →
X 
GitHub Background
Shipper co-founded Firefly while a student at the University of Pennsylvania. Penn’s account identifies him as a College alumnus of the class of 2014 and explains Firefly’s co-browsing technology: a remote participant could connect to another person’s browser window and collaborate without a separate download. His personal biography dates the sale to Pegasystems to 2014. The product addressed customer-support work where seeing the same browser state can be more useful than describing a problem over text.[1][4]
Career
At Every, Shipper combines writing and interviews with product development. His personal site describes a company that publishes a daily newsletter, develops applications and trains leadership teams in workplace AI. The listed products include email, dictation, file organization and collaborative editing tools. These are company products; his sourced personal contributions here are leadership, his Chain of Thought column and the interview show he co-hosts.[1][2]
Ideas
In his January 2024 essay, Shipper argues that delegating tasks to AI changes the skills workers need. He starts from his own use of ChatGPT for summaries and proposes a shift toward choosing tasks, assigning resources and evaluating the results. He calls that an allocation economy, extending management-like responsibilities to individual contributors. It is an argument about how work may change, grounded in his own workflow observations, rather than a measured claim that the entire economy has already made that transition.[3]
Tools, writing & talks
Writing, research & talks
Start here
An argument that AI delegation shifts human work toward selecting tasks and evaluating results.
Start with the allocation-economy essay to understand the idea behind his AI writing. It needs no technical background; compare its delegation argument with a task you currently perform yourself.
Author[3]Every's Dan Shipper reports on Anthropic's Duet, an always-on personal agent in its own iOS, Android and web app, with channels and a todo list. It reads local Claude sessions but cannot direct them, and voice is transcription only. Includes a nine-agent comparison page.
Catalog workOriginal sourceEvery's reported history of Codex: a failed cloud launch, a rebuilt local agent in Rust, a separate coding model line, a desktop app without an editor, and a merge into ChatGPT. It shows how product form factors tracked model leaps.
Catalog workOriginal source- Senior engineer benchmark: eight models, one refactor taskchecks.every.to
Every's Senior Engineer Benchmark scores eight models across Claude Code, Codex CLI and Muse Code on one codebase-rewrite task. Scores run from 85% to 21%. Single task, one author's checks, LLM-judged by Claude.
Catalog workOriginal source Every's week-long test of Google DeepMind's Gemini 4 Argon. It matches Opus 5.5 on a greenfield build and is strong for writing, but weak on existing codebases and self-review, and Antigravity lags Claude Code and Codex. Based on the thread summary; the full piece is paywalled.
Catalog workOriginal source- Chain of Thoughtevery.to
Shipper’s essays on using AI, cognition and the organization of work.
Author[2] - Firefly’s student-founder historymagazine.wharton.upenn.edu
Penn’s account of the co-browsing startup and its early funding.
Profile subject and co-founder, not article author[4]