Every trained an AI agent on 30 000 edits by editor-in-chief Kate Lee
Every trained an AI agent for copy-editing on 30 000 historical edits by its editor-in-chief Kate Lee and tested it retrospectively against her older texts — according to CEO Dan Shipper, the goal is to distribute one person's expertise throughout the company.
Every, a company that runs a publication about AI and also develops products built on language models, has created an AI agent for copy-editing, according to its co-founder and CEO Dan Shipper. The agent was trained on a dataset of 30 000 historical edits made by the company's editor-in-chief Kate Lee, and was then tested retrospectively against her older texts. According to Shipper, the goal is to capture the expertise of one specific person and distribute it throughout the organization.
According to the article, Every has roughly 30 employees and combines journalism (the Chain of Thought column, the AI & I podcast, and reviews of new models called “vibe checks”) with product development — the email assistant Cora, the file organizer Sparkle, the writing tool Spiral and the dictation app Monologue, which are offered alongside the content in a subscription costing 20 dollars per month. According to Shipper, AI currently writes practically all of the company's code, while people continue to do the journalistic work.
The source mentions that Every also publishes critical reviews of models, including models from Anthropic, whose technology it also uses in part. According to Shipper, the role of independent evaluator is one of the company's most enduring sources of value, because “no one trusts any lab to give an objective assessment of its own model”.
The remaining part of the interview, including further details about the company's increase in employee numbers despite automation, is not available in the source. Details can be found in the source article.
Why it matters
The case shows a concrete method for a company to capture the tacit know-how of an experienced employee (their style and decision-making when editing) and turn it into a reusable tool for the whole team, without layoffs necessarily being required — according to the article, Every doubled its employee count over the past year. For editors and other knowledge workers, this signals that their expertise can be a source of training data, changing the nature of their role in the organization.
Two audiences, two different impacts
What this means
For individuals
For editors, journalists and other knowledge workers, the case shows that their individual style and decision-making can be translated from historical data into a tool used by a wider team — changing the expert's role from someone who carries out the work to a source of training data.
For a business
Companies may consider a similar approach — capturing a key expert's decision-making patterns in historical data and using them to train an internal tool that distributes their expertise throughout the organization, without necessarily replacing the employee.
ProcessesCheck the original
Event sources
only one source so far · 1 publisher, 1 independent. We count feeds from the same owner only once.