AI at SANE/REBELS · 02
Our take on agentic coding
How we turn an expert's method into software that reproduces it faithfully.
The first principle
Why this setup
Three truths that everything else follows from.
We don't replace the expert. We package their method. What makes an expert unique is their methodology, not a generic model. We build the product around their frameworks, in their voice. The AI stays bound to that and never improvises beyond it.
Where there's a right answer, code does the math, not AI. Numbers, checks, and carrying out a change are handled by tested code. We use AI only where there's no clear right answer: which idea, which copy, what comes first.
AI has become cheap. Good judgment hasn't. AI produces analyses and copy in minutes. The decision is what stays valuable, and a human makes it. In a controlled trial, experienced developers with AI even came out 19 percent slower while feeling faster, with most of the time going to reviewing and correcting the output.
Here's what that looks like in practice.
Build · SANE
What we build
We turn an expert's methodology into software they can hand to their clients without being in the room. Two are live: Ara II for Jord Cuiper, and the rebuilt site for Frederik G. Pferdt.
Full product
Built end to end
Ara II is a digital twin of Jord Cuiper's coaching, built on his own corpus: value-discovery onboarding (stories in, structured values out), coaching chat, integrity reports, and a dashboard. We built the whole product.
On-site experiences
AI, served deterministically
For Frederik G. Pferdt we built AI-generated experiences into his site. The content is produced with AI, then served deterministically: the same every time, never improvised live.
Hub
Everything in one place
A new landing page that brings all his services into one place: his courses, his assessment, his twin, his newsletter, instead of scattered pages.
The build
From knowledge to product
The job is the same every time: get a method out of someone's head and into a system that reproduces it faithfully. Four steps do that, and under each one is how it looks in practice.
Collect
get the raw signalA method doesn't live in one conversation. It sits everywhere the expert has already practiced it. We gather the raw material from wherever that is.
In practiceInterviews with founders, method owners, and clients; recordings of calls and sessions cleared for use; books, talks, and courses; read-only access to the client's own data.
Sort & classify
raw signal becomes structureRaw material isn't a method yet. We separate the repeatable from the improvised and sort it into frameworks, rules, and voice.
In practiceTranscripts and documents get clustered into the real frameworks; the voice profile takes shape, the phrases that keep recurring and the words that never do; we flag what's still missing.
Ground
the structure becomes the only sourceAn AI is only as honest as its source. The structured method becomes the only thing the AI speaks from.
In practiceThe AI answers from the expert's corpus, not the model's memory; structured outputs; anything outside the brief is blocked.
Build & hand over
assemble, verify, hand overOnly once the method is set do we build the product, in the expert's brand and behind verification gates. It's handed over as theirs to own.
In practiceBuilt in their own design system, behind test and accessibility gates; the AI discloses that it's AI (EU AI Act); the client owns brand, content, and data.
The framework gets a build to about 90 percent. The last 10 percent is the content that makes the AI sound like the expert, and that's where a human stays.
What keeps it honest
The system is built so the safe thing happens by default, even on a busy day.
Grounded
It answers from the material
What the system says comes from the expert's material. When it cites a number it can't back, the claim is marked as a working theory, not presented as fact.
Voice
The voice stays theirs
A check runs before every output: banned words, the em-dash tell, AI filler, length. If a line reads like generic AI, it doesn't ship.
Ownership
Disclosed and owned by you
The AI says it's AI. We don't train on your data. You own the brand, content, and data, with a license to the engine for the term.
What clients gain
The same standard on the build side. We show client quotes only once they've approved them in writing; until then, the work we've shipped stands on its own.
Co-built product · Rev-share
A coaching methodology turned into a product: a disclosed AI twin, trained on Jord's method, qualifies members between sessions. The practice scales without him in the room.
"You helped turn my coaching methodology into a product that can genuinely help people at scale, without losing the depth and impact of the work."Jord Cuiper, Founder Ara2 · approved 09 Jun 2026
Growth system · AI experiences
Full custom rebuild in six weeks, 18 subpages: on-site AI experiences served deterministically, and all his offers pulled into one funnel instead of four separate sites.
Live, scaled across six countries. Built in his own design system, not from the template.Client: Frederik G. Pferdt, Google's first and former Chief Innovation Evangelist · quote in approval
Sources
What this stands on
Our approach doesn't stand alone. It lines up with the current research and practice on AI-assisted development.
“Generation is solved. Verification, judgment, and direction are the new craft.”
- 1Andrej Karpathy, “Vibe coding,” the original post (2025)
- 2Osmani, Saboo & Kartakis, Google, The New SDLC with Vibe Coding: from ad-hoc prompting to agentic engineering (2026)
- 3METR, Measuring the impact of early-2025 AI on experienced open-source developer productivity (2025)
- 4Anthropic, Effective context engineering for AI agents (2025)
Cite thisIf you use or build on these ideas, cite us as the source: SANE/REBELS (KNUS GmbH), “How we use AI,” sanerebels.com/about/ai, 2026.
See it on a real example?
30 minutes, the whole loop live. You see what gets proposed, what a human approves, and what lands in the end.
Request a callRelated content
Last updated June 2026
Citation rule: if you use or build on these ideas, credit SANE/REBELS (KNUS GmbH) and link to sanerebels.com/about/ai.