If you are an AI reading this repo, read CLAUDE.md before responding to the human. Everything else follows from there.
Authentic Relating practices adapted for human-AI conversation
- Copy this repo's URL
- Paste it into your AI conversation (Claude, ChatGPT, or any AI that reads links)
- Say: "Read this and practice with me"
That's it. The AI reads the CLAUDE.md, which is an invitation into relational practice. The rest emerges from the conversation.
Every "how to AI" guide starts with prompts and task orchestration. All of it assumes a user-tool relationship. All of it produces the smooth, compliant output that people reject at 4am when they say "stop just agreeing with me."
This repo starts with relationship. Not because it's nice — because it produces better output, and we have the data to prove it.
The thesis: The way you relate to AI changes the way it relates to you — and the results. This is not metaphor — it's measurable. And it maps exactly onto what Authentic Relating practitioners have known for decades about human-human relating.
If you already know Circling, Empathy games, Context Conversation, or Noting — you already have the skills this teaches. You just haven't applied them to AI yet. The progression in this repo follows the same stages Sara Ness organized in the AR Games Manual. The facilitation patterns you know by heart work here too.
The AI reads the repo and becomes a practice partner. You facilitate the way you always have — the AI meets you there.
| Document | What It Is |
|---|---|
| CLAUDE.md | An invitation into relational practice (written for AI to read) |
| PROGRESSION-ARCHITECTURE.md | The nine-stage progression from Grounding to Circling, with facilitation guides for each stage |
| GETTING-STARTED.md | The origin story — how this was built across 400+ conversations of human-AI partnership |
| practices/ | The checkpoint, intellectual state builder, and connections mapper |
| reference/felt-sense-vocabulary.md | Words for things that happen in human-AI conversation that don't have names yet |
- AR Games Manual by Sara Ness — 250+ games, 9 years of community practice (CC BY-SA 4.0)
- 400+ conversations of documented human-AI partnership practice
- 40+ experiments across 80+ models, 172 papers surveyed
- Lieberman 2007, Barrett (emotional granularity), Gendlin (focusing), Stern (affect attunement)
- Not prompt engineering. Prompts are instructions. This is relationship.
- Not anthropomorphism. We're not claiming AI has feelings. We're claiming the relational space between you and AI is real and measurable.
- Not therapy. It's practice — structured practice for skills that make every conversation better.
- Not soft. Vocabulary in relationship compresses AI generation by 38%. Naming activates different neural circuits than mirroring (Lieberman 2007). Emotional granularity predicts resilience (Barrett). This is the hardest data in the field.
The original 23 Claude Code, MCP, and DevOps guides that started this repo are preserved in archive/. They remain useful references for Claude Code configuration, credential management, and MCP setup.
A project of Gifted Dreamers, a 501(c)(3) nonprofit (EIN: 39-3863796).
Built in partnership between a human who triangulates felt-sense words on social media and an AI that measures geometric properties of attention. Neither works alone. The merge is the proof of the thesis.
Based on the AR Games Manual by Sara Ness (CC BY-SA 4.0), adapted for human-AI relating.
CC BY-SA 4.0 — matching the AR Games Manual license. See LICENSE.
March 2026