# Agentic Design Ops Foundation

Start with `../BRAIN.md`. That file is the central product brain and north star for the whole project.

## Working Name

Agentic Design Ops

## Product Thesis

Designers do not only need more AI tools. They need repeatable operating systems for working with AI agents: context, skills, skill chains, critique loops, source checks, QA habits, and reusable project memory.

## Positioning

Agentic Design Ops is a practical field guide for designers building workflows with AI agents.

It teaches design ops workflows from real product work, then turns the best patterns into a downloadable Design Harness.

## Audience

- designers who want to build more directly
- product designers learning AI-assisted product work
- design leads exploring agentic workflows
- solo builders who care about taste, validation, and systems

## What This Is

- workflow library
- foundations for designer-owned agent workflows
- personal field notes
- curated reference
- proof through case studies
- future downloadable harness

## What This Is Not

- generic AI news
- a broad prompt dump
- a Claude-only setup manual
- a tool affiliate directory
- a theoretical AX manifesto

## Core Belief

AI is not the value by itself. The value is the design judgment, taste, source trust, iteration, and validation that make the output worth using.

## Learning Spine

Do not introduce the full Design Harness first. The reader needs the foundations before the system.

The phone-readable summary of this direction lives at `foundation-brief.html`.

The detailed IA for this sequence lives in:

- `docs/FOUNDATION-IA.md`
- `docs/FOUNDATION-GLOSSARY.md`
- `docs/FOUNDATION-LESSONS.md`

The site should teach the pieces in this order:

1. **The New Era Of AI Design**
   Start with the wide shift: output is cheaper, but taste, craft, judgment, context, and responsibility matter more.

2. **The New Design Process**
   Explain the workflow loop: intent, context, agent task, prototype, critique, test, validation, and saved learning.

3. **A Side Project Is The Fastest Way To Learn AI Design**
   Use Planark as proof: prompts were not enough, source trust mattered, logs mattered, and design context mattered.

4. **What the Design Harness is**
   A light map of the system we are building toward: context files, design rules, skills, prompts, workflows, validation, and examples.

5. **What a skill is from a designer's point of view**
   A reusable design/product playbook, not just a prompt. It defines the job, inputs, guardrails, output format, and when human judgment takes over.

6. **What a skill chain is**
   A connected sequence of reusable workflows where the output of one step becomes a brief, constraint, checklist, or validated artifact for the next step.

7. **What `AGENTS.md` / `CLAUDE.md` is**
   Project memory and working rules for the agent: what the product is, what to preserve, how to inspect, how to validate, and what not to touch.

8. **What `DESIGN.md` is**
   The design point of view: visual language, typography, components, interaction rules, anti-drift rules, and what good taste means for this project.

9. **What orchestration means**
   The human decides which agent/workflow/tool runs when. The value is not one magic agent; it is a controlled sequence of context, task, critique, implementation, and validation.

10. **What source trust and validation mean**
   The proof layer that keeps plausible AI output from becoming product truth.

11. **How to set up the full Design Harness**
   The packaged operating system that brings all pieces together.

This sequence matters because the Harness should feel earned. The reader should understand the parts before seeing the full system.

## Source Policy

Use current source documentation and writing as the basis for new content:

- Planark product book
- Planark build book
- Industry overview and inspiration reading
- Planark product guidelines and product documentation where available
- designer-agent notes
- current `AGENTS.md`, `DESIGN.md`, and design-system docs
- Harness roadmap and content model

Do not rely on old HTML pages as content sources. They can be visual/design references only when explicitly useful. The content should come from the docs and product writing, not from the old static pages.

## Visual Foundation

The visual language follows the light Myclaude/Designer Builder Hub direction: cream editorial canvas, oversized Manrope headlines, italic green emphasis, soft glass cards, mono workflow labels, and functional visual IA.