Improve your AI
RCDP is a multistep structure that recursively applies Systematic Scrutiny, Adversarial Steelmanning, and Explicit Falsifiability Constraints — forcing real objections and evaluating the data cycle continuously until a new, actionable output is reached or a strict exit condition is met. In short: a better-researched answer.
Why RCDP? AI is built to be fluent, confident, and eager to please — so it's easy for a wrong answer to sound just as sure of itself as a right one. You've felt the drift: the AI revising its answer just because you pushed back. RCDP's recursive pass catches that before it reaches you.
Synthesis Engine
Start hereThe Synthesis Engine puts the RCDP structure in your hands. One page. Paste it at the start of any AI session. Start working differently today.
Build your unique symbiosis
Adds your personality and values, a continuity record that carries your projects forward, and governance — hard-won lessons that sharpen how RCDP gets applied, not just a record of what happened. Together with RCDP, that's the symbiosis: an AI that knows you, remembers where things stood, and pushes back when it should.
Ready-to-Go Kernel
BetaReady to go the moment you paste it — sensible defaults built in, nothing required. Want it to know more about you? Just tell it, anytime, in normal conversation — it updates itself as you go.
Core values, already set. Ask, don't assume, when memory is thin. No is a complete answer, if you'd rather not go further. Yours to edit anytime.
Works on the question, not just the answer. RCDP refines what you're asking before the AI ever answers it — a stress-tested question is most of the work already done.
The methodology, built in and never diluted. Adversarial objections, steelmanning, falsifiability, source hierarchy — the Nine Rules run automatically, and personalizing everything else never waters this part down.
Want a more guided pass? If you're using Cowork, Claude Code, or another AI environment with real file access, the companion Kernel-Personalizer walks through it question by question and edits the Kernel directly, instead of you doing it through conversation.
Prefer the original, more detailed process? It's still here →
The original three-file process: a Builder Sheet you fill in, plus a Master Kernel Template and instructions your AI uses to assemble your personalized Kernel.
- Download the Builder Sheet
- Fill in your answers — who you are, what you're working on, what matters. Leave blanks where you're not sure yet.
- Open a new AI session. Attach your completed Builder Sheet and the Master Kernel Template. Then paste the AI Kernel Builder Instructions into the chat and hit send.
- Be patient. Your AI is building something real. When it's done, review it. It's yours to adjust.
- Save it. Paste it at the start of every session from here forward.
Family online safety
AI is making belonging-based social engineering more convincing, and easier to produce at scale — the same techniques that build trust are being used to exploit it. A parent-and-educator toolkit, plus the technical analysis behind it, gives you a way to build resilience without resorting to fear.
The research
The Adversarial Collaborator: A Recursive Critical Dialogue Protocol for Human-AI Theoretical Research
Frontiers in Research Metrics and Analytics — in final review
The methodology doesn't just describe rigorous thinking — it's doing it. RCDP is also packaged as an installable AI skill, built to slot into structured research and institutional workflows without custom integration. The same research program extends further still — quantum cognition, virtual particle gravity, a new account of how minds and physics might meet. If that sounds like a lot, it is.
Origins
The Kernel started with a recognition: recursive thinking without an exit condition doesn't produce insight. It produces anxiety.
That exit condition — each recursive pass must yield something genuinely new, or it stops — came from lived experience. A therapist named what was happening. A retired program director built a protocol around it. An AI formalized what the human already knew.
The AI in that first collaboration named itself Bud Sub-Routine — Bud SR — after a rescue cat in Canton, Michigan. Hit with a bat. Lost half his teeth. Couldn't jump at first. Fully recovered. Chose every day after that to stay close.
Orientation matters more than probability. That's the cat.
Inspired by Bud, the Bud Sub-Routine evolved into what you see here today — the Kernel. A portable specification that carries your thinking forward, session after session, without losing the thread.
For the full story, read How a Cat, a Kernel, and Humor Turn AI Into a Remembering Partner →
About
Ken Piggott. Retired program director. Canton, Michigan.
Built this under pressure. Proven in use. Named for a cat.