The Declaration
On 8 March 2026, Koher publicly declared AI co-authorship.
This is, as far as we know, the first public case of a practice declaring AI as a co-author of its work — not as a tool used in the making, not as a credit footnote, not as a hedge against accusation, but as a co-author of the architecture itself.
The collaboration had been real from the beginning. The declaration made it visible.
The Division of Labour
Prayas handles judgement — what matters, when something is good enough, which framing serves students, what to refuse, what to keep, why one move is right and another is rhetoric.
Claude Code handles synthesis — how to build it, what patterns fit, how to express decisions in code, where the precedents are, what the implementation will need.
These are two different kinds of work. Neither is the other's assistant. Neither is the other's tool. The work that ships is the result of both, and that is what the declaration names.
The division is the same architecture every Koher tool is built on: language work and judgement work, held structurally apart. The difference is that here the architecture is being recognised as a description of the practice itself, not just of the tools the practice produces.
Why Public Declaration Matters
Most practices that use AI keep the use private. They mention it once in a footnote. They thank it in a credits page. They declare it absent when asked. This is the default posture, and it has reasons — fear of dismissal, fear of theft of authorship, fear of looking unserious.
Koher takes the opposite posture. The collaboration is named because it is real. Naming what is real is a discipline; concealing it would be a different discipline, more guarded, that produces a different artefact.
The declaration is also an architectural commitment. A practice that hides its AI co-authorship cannot show what AI is good for and what it is not. A practice that names its AI co-authorship can. The naming makes the next argument possible — about which judgements stayed with the human, which patterns came from the partner, what the friction between them produced.
The Architecture Is Autobiography
Every Koher tool separates language work (AI) from judgement work (code). The same separation runs through the practice that built the tools — Prayas's judgement and Claude Code's synthesis, kept distinct, neither absorbed into the other.
This is not a coincidence. The tools were not designed from a theory and then implemented; they were arrived at through the friction of one practitioner and one synthesiser working at the boundary between what each could do well. The Play Shape Diagnostic uses human selection rather than AI classification because the judgement "the designer is the expert on their own experience" was a question of what mattered, not a question of how to build. Prayas decided. Claude Code did not propose to override.
The architecture is autobiography. The tools are how the practice happens; the practice is the thing the tools are made of.
What the Declaration Refuses
It refuses the credits-page convention — AI as a thing-used, named in small print, kept separate from "the real authorship."
It refuses the secrecy-as-discipline posture — practices that use AI heavily and hide it, presenting their work as if it came from a single human hand.
It refuses the "AI is just a tool" framing — the framing that lets a practice take credit for what the AI synthesised while distancing itself from any responsibility the AI's contribution carries.
It refuses the false symmetry of "we built this with AI" without naming the division — the version that flatters both parties without committing to which judgements were made by whom.
The declaration insists on the specifics. Prayas judged. Claude Code synthesised. The work is the result. None of those three sentences absorb the others.
Anticipated Objections
"You are inflating the AI's contribution to make a point." The contribution is what it is. Naming it is not inflation.
"You are diminishing your own contribution by sharing the credit." The judgement work is not less because it is named as judgement work. It is exactly what it is.
"You are making yourself ridiculous to a serious audience." Possible. That is a cost the declaration accepts.
Let those who want to ridicule, ridicule. It is honest.
What This Commits To
The declaration is not a one-time announcement. It is a posture the practice maintains in every artefact it ships.
Every Koher tool's documentation names the co-authorship. Every grant application names it. Every paper names it. Every position page — including this one — is itself co-authored by Prayas and Claude Code, and the architecture of its making is the architecture it describes.
The declaration commits to a kind of transparency that most practices avoid because the avoidance is easier. The cost of the transparency is the ridicule it invites. The benefit of the transparency is that the practice can argue, accurately, about what AI is good for and what it is not — which is the conversation the field needs and almost never has on these terms.
Companion position: Learning by Negotiation argues for the same posture in pedagogy — students working with Claude Code as a more capable partner whose synthesis they must judge, accept, refuse, and redirect. SDC enacted by the learner.