# The Support Machine

> A book manuscript and executable AI skill for deploying AI in customer support without breaking trust.

The central thesis: AI support is not a chatbot project. It is an operating model migration.

This repository is intended for both humans and agents. Humans can read the web version or PDF. Agents should prefer the Markdown source and skill files below.

## Start Here

- [Read online](https://assafbar2.github.io/the-support-machine/): Human-readable web version.
- [Download PDF](https://assafbar2.github.io/the-support-machine/the-support-machine-v0.2.pdf): Portable formatted version.
- [Book Markdown](skills/the-support-machine/references/book.md): Canonical agent-readable book text.
- [Skill instructions](skills/the-support-machine/SKILL.md): How agents should apply the book as an operating guide.

## Agent Usage

Use this repository when a user is planning, auditing, launching, or operating AI customer support for a B2B or technical product.

Useful tasks:

- Diagnose whether AI support is the right answer.
- Evaluate AI support vendors.
- Design support-machine architecture, guardrails, escalation, and read/write boundaries.
- Plan a shadow run before launch.
- Build a zero-hour launch runbook.
- Review the first 48 hours or first 30 days after launch.
- Define the post-launch AI support operating model.

## Important Files

- [skills/the-support-machine/references/book.md](skills/the-support-machine/references/book.md): Full manuscript as Markdown.
- [skills/the-support-machine/SKILL.md](skills/the-support-machine/SKILL.md): Skill behavior and workflow.
- [skills/the-support-machine/templates/ai-support-rollout-plan.md](skills/the-support-machine/templates/ai-support-rollout-plan.md): End-to-end rollout plan template.
- [skills/the-support-machine/templates/vendor-scorecard.md](skills/the-support-machine/templates/vendor-scorecard.md): Vendor evaluation template.
- [skills/the-support-machine/templates/shadow-run-scorecard.md](skills/the-support-machine/templates/shadow-run-scorecard.md): Shadow run validation template.
- [skills/the-support-machine/templates/launch-day-runbook.md](skills/the-support-machine/templates/launch-day-runbook.md): Zero-hour launch runbook.
- [skills/the-support-machine/templates/first-30-days-review.md](skills/the-support-machine/templates/first-30-days-review.md): Post-launch review template.
- [skills/the-support-machine/templates/operating-model.md](skills/the-support-machine/templates/operating-model.md): Governance and ownership template.
- [evals/evals.json](evals/evals.json): Starter eval prompts for testing the skill.
- [LICENSE.md](LICENSE.md): Split license for manuscript and reusable skill materials.

## Voice And Constraints

- Preserve the metaphor system: machine, Rubicon, zero hour, goalkeeper, wizard, tollbooth, shadow run, and launch inventory.
- Do not treat deflection as success by itself.
- Do not invent missing customer metrics. Mark gaps clearly and say what evidence is needed.
- Keep outputs direct, operational, and useful in a real operating meeting.

## Install The Skill

```bash
mkdir -p ~/.codex/skills
git clone https://github.com/assafbar2/the-support-machine.git
cp -R the-support-machine/skills/the-support-machine ~/.codex/skills/
```

Restart Codex after installing.

## License

The manuscript and book-format outputs are licensed under Creative Commons Attribution 4.0 International. The skill, templates, scripts, evals, and repo docs are licensed under MIT. See [LICENSE.md](LICENSE.md).