Kenn So — Generational - Agents at Work - September 2025

Free

Deeply researched analysis on the business technology of AI

FreeFree tier

Type

Open Source

Visit Website GitHub (38 stars)

About Kenn So — Generational - Agents at Work - September 2025

Generational is a Substack publication by Kenn So that provides deeply researched, data-driven analysis on the business technology of AI. It covers topics such as AI agent market maps, occupation-level automation potential, and investment strategies. The specific article 'Agents at Work – September 2025' presents a framework for locating where AI agents create value, based on 19,000 tasks, 830 occupations, and wage pools totaling $10 trillion. The analysis incorporates Anthropic's Economic Index of 1 million conversations to map usage and capability, identifying high-potential areas like software development, customer service, and accounting. The publication also offers essays, company briefings, and trend analyses aimed at helping investors, operators, and decision-makers.

Key Features

Data-driven framework for AI agent market mapping

Dataset of 19,000 tasks and 830 occupations with wage pools

Integration of Anthropic Economic Index usage data (1 million conversations)

Occupation-level analysis of agentic automation potential vs. adoption

Publicly available slides, data, and methodology

Regular essays, company briefings, and trend reports (1-3 per quarter)

Pros & Cons

Pros

Empirical, data-driven approach using real wage and usage data
Covers a broad range of occupations (830) for granular insights
Updated regularly with new data (expanded from previous year)
Provides actionable signals for where agents are taking entry-level tasks
Free access to all content and underlying data

Cons

Analysis is at the occupation level, may not capture specific task nuances
Usage data is aggregated at major-occupation groups, requiring further detail for practical targeting
Not a product or tool, but a research resource requiring interpretation
Publication frequency is low (1-3 essays per quarter)

Best For

Identifying high-potential occupations for AI agent deploymentMarket sizing and investment decisions in AI startupsUnderstanding where AI agents are already adopted in productionStrategic planning for AI product development and go-to-marketMonitoring employment impact of AI across age cohorts and occupations

FAQ

What is Generational?

Generational is a Substack publication by Kenn So that provides deeply researched analysis on the business technology of AI, aimed at helping readers make decisions about investments, business strategies, product decisions, and company choices.

How often is content published?

The publication publishes 1-3 essays per quarter, along with occasional company briefings and trend reports, only when the content is considered worth the reader's time.

What data does the 'Agents at Work' article use?

The article uses a dataset of 19,000 tasks across 830 occupations, wage pools from labor data, and usage data from Anthropic's Economic Index of 1 million conversations to map agentic automation potential and observed adoption.

Is the underlying data and methodology available?

Yes, the article provides links to slides, data, and the methodology used for the analysis.