# Case Studies

This lane turns abstract ideas into bounded, story-led examples that connect
architecture choices to recognizable agent products and workflows.

## What belongs here

- Deep research agents
- Coding agents
- Customer support agents
- Research assistants
- Travel and concierge agents
- Document-analysis agents
- Media and content workflows

## Editorial intent

Each case study should start with a user problem, then move through system
shape, tool choices, memory strategy, evaluation questions, and likely failure
modes.

## Current pages

- [Deep Research Agents](./deep-research-agents.mdx): a bounded look at
  planning, evidence collection, synthesis, and citation discipline.
- [Coding Agents](./coding-agents.mdx): a repository-first case study for
  scoped edits, verification, approvals, and human review.
- [Customer Support Agents](./customer-support-agents.mdx): a local-first case
  study for email triage and policy-grounded reply drafting.

## Example starters

- [Deep Research Agent Starter](./examples/deep-research-agent-starter/index.mdx):
  a small project scaffold for planning, evidence collection, and cited output.
- [Customer Support Email Agent Starter](./examples/customer-support-email-agent-starter/index.mdx):
  a small local-document workflow for triaging customer emails and drafting
  policy-grounded replies.
- [Customer Email Assist Starter](./examples/customer-email-assist-starter/index.mdx):
  a mailbox-integrated dashboard starter for Gmail sync, SQLite issue queues,
  deterministic review flows, and policy-grounded replies.