Factory

Overview

Factory is an AI-native startup highlighted as an example of an AI-first operating model, where work is increasingly delegated to AI agents across functions such as engineering, product management, design, and sales. In the newsletter mentions, Factory appears not just as a company building AI tools, but as one actively organizing work around agents, workflows, specs, and feedback loops.

For AI Product Managers, Factory matters because it represents a practical model for how teams may operate as AI capabilities mature: humans provide direction, context, judgment, and system design, while agents handle substantial execution. Its ecosystem—including Droid, Ghosty CLI, and Eno’s PM skill—shows how agentic workflows can support both product strategy and hands-on delivery.

Key Developments

• 2026-02-14: Factory is cited alongside Ramp and Linear as an AI-native startup delegating tasks to AI agents across engineering, PM, design, and sales, with humans focusing on context, systems, and feedback loops.
• 2026-02-15: Peter Yang previews Eno’s new AI-powered product management skill built at Factory, describing a 700+ word system covering product principles, positioning, 11-star experience guidelines, PRD templates, review rubrics, and prioritization frameworks.
• 2026-02-16: Peter Yang demonstrates Factory’s Droid agent through the Ghosty CLI in high-autonomy spec mode, using Opus 4.5 for planning and GPT-5.2 for execution to build and QA a React speed-reading web app. The workflow includes spec generation in VS Code, Chrome DevTools screenshots, linting, and type-checking.

Relevance to AI PMs

1. A concrete model for AI-first operating design Factory shows how PMs can rethink team structure around agents rather than treating AI as a lightweight assistant. The tactical takeaway is to define what humans own—context, prioritization, quality bars, and feedback loops—versus what agents can execute repeatedly.

2. Spec-driven agent workflows for product delivery
The Droid + Ghosty CLI example is useful for PMs who want to operationalize spec-based execution. Instead of starting from prompts alone, PMs can use structured specs, editable planning documents, and explicit execution phases to improve reliability.

3. Codifying PM judgment into reusable skills
Eno’s PM skill suggests that product thinking can be turned into reusable AI instructions, templates, and rubrics. PMs can apply this by encoding their own PRD formats, review criteria, positioning frameworks, and prioritization logic into internal AI systems.

Related

• droid: Factory’s AI agent used for coding and product work in a high-autonomy workflow.
• ghosty-cli: The command-line interface used to operate Droid, including spec mode and execution flows.
• opus-45: The planning model used in the documented Droid workflow.
• gpt-52: The execution model used alongside Opus 4.5 in the Factory agent workflow.
• eno: Associated with Factory through an AI-powered product management skill containing PM frameworks and templates.
• ramp: Mentioned with Factory as another AI-native company delegating work to agents.
• linear: Mentioned alongside Factory and Ramp as part of the AI-native operating model discussion.
• peter-yang: The newsletter source who highlighted Factory’s workflows, tooling, and relevance for product work.