I hadn't announced it here yet: I am writing a book about Contexte.

The cover of the book being written.

Contexte is the method I am building to let a human and an AI work durably around a mission, without starting over at every conversation, while keeping:

• the sources;
• the knowledge produced;
• the decisions;
• the state of the work and what will make it possible to pick it up later.

The book is now almost finished.

At this stage, I could have kept reading it through, fixing a few formulations, filling out the examples and preparing for publication.

Instead, I asked myself a fairly uncomfortable question:

what did I really invent in there?

Not "invent" in the sense of having to claim every idea.

The question was much simpler:

am I presenting as a new method something that already exists and that I have simply reinvented without knowing it?

This is above all a question of scientific culture of doubt.

When you build something for months, it is very easy to unconsciously look for everything that confirms your idea is good, different or original.

I tried to do the opposite.

To methodically look for everything that could take that conviction apart.

In everything I present as the Contexte method: - what already existed? - what did I consciously borrow? - what did I adapt? - what did I rediscover without knowing it? - what had already been assembled elsewhere? - and what, if anything, remains specific to the way I brought it all together?

At the start, I thought I was simply doing a prior-art search.

In practice, what I mostly learned was to take my own certainties apart, step by step.

I already knew I hadn't started from nothing

Some influences were obvious.

The bounded context from Domain-Driven Design is an acknowledged inspiration.

Atomic notes inevitably led me to Zettelkasten.

PARA and the Second Brain shaped the way I think about knowledge spaces.

On the more technical side, I had also consciously reused intuitions coming from Git — around separation, copying and merging in particular — as well as the Observer pattern for certain relations between Contexts.

So the question was not:

"Are all my ideas new?"

I already knew the answer was no.

What interested me was the assembly.

A Contexte, in the book, is a space bounded by a mission. It holds sources, knowledge produced from those sources, a current state, a history, decisions, contradictions, actions, deliverables. It has to be interruptible and resumable. It has to accumulate knowledge as the work goes on. And several Contexts have to be able to exchange without losing their autonomy.

So I told myself: the building blocks surely come from elsewhere.

But maybe their combination is distinctive.

That is the hypothesis I set out to knock down.

First mistake: searching with my own words

The first problem appears immediately when you look for the prior art of a method.

Nobody is obliged to have used your vocabulary.

Searching "Contexte" + AI + memory tells you almost nothing about whether a similar mechanism exists.

A medical team may have been doing something resembling your handoff for fifty years, without ever using that word.

An archival system may have formalized provenance, source and interpretation without ever mentioning knowledge management.

A piece of software from the 1990s may have solved a copying and dependency problem you think you ran into with AI agents in 2026.

So I had to stop searching for Contexte.

I started by taking the book apart.

With ChatGPT, we extracted 24 concepts and 12 structuring relations.

The idea was to turn the book into objects simple enough to ask questions about.

Not only:

Has anyone already imagined a persistent workspace?

But also:

Has anyone already separated the raw source from the derived knowledge?

Does any system distinguish the present from the historical log?

Is there a tradition where work on one problem enriches a base reusable for the next ones?

What happens when a piece of knowledge is copied from one autonomous space into another?

That changes the search completely.

First surprise: almost all the building blocks were already there

Once the system was taken apart, antecedents started appearing everywhere.

A few examples:

• provenance obviously had a whole literature behind it, right up to standards like W3C PROV; W3C — PROV-O: The PROV Ontology
• Event Sourcing offers a way to rebuild a current state from the history of events;
• Architecture Decision Records keep a decision, its context and its rationale, including once it has been superseded; Michael Nygard — Documenting Architecture Decisions
• Truth Maintenance Systems were already working, back in the 1980s, with several potentially incompatible sets of assumptions. Johan de Kleer — An assumption-based TMS

At that stage, none of this really surprised me.

A complex method is rarely made of mechanisms that came out of nowhere.

The first conclusion looked fairly comfortable:

the building blocks are not new, but the assembly may be.

It was a reasonable conclusion.

It did not hold up for long.

Second iteration: search for the function, not the concept

We changed method.

Instead of searching for the concepts in the book, we started searching for the problems they answer.

• How do you pick up work again after an interruption?
• How do you hand a case over well enough for someone else to carry on?
• How do you turn thousands of pieces of information into evidence, then into hypotheses?
• How do you keep several competing explanations without artificially picking one?
• How does a workspace hold both its operational state and reusable knowledge at the same time?

This time, the search left software very quickly.

It took us towards: - intelligence analysis; - law; - medicine; - incident management; - cognitive science; - archival science; - scientific research.

The sensemaking model of Peter Pirolli and Stuart Card, for instance, describes a process in which the analyst starts from external sources, selects the relevant information, builds up items of evidence, constructs schemas and hypotheses, then produces the result of the analysis. Pirolli & Card — The sensemaking process and leverage points for analyst technology

This is obviously not Contexte.

But the proximity of the mechanism was interesting.

Same thing with Jakob Bardram's Activity-Based Computing.

In 2005, Bardram describes an infrastructure where the central unit is no longer the application but human activity, with the associated computing tasks and resources. The architecture is designed in particular to support mobility, collaboration and the continuity of activity across environments. Aarhus University — Activity-Based Computing: Support for Mobility and Collaboration

In other words, the idea that you have to keep something richer than a document in order to make work continuous had obviously not waited for AI agents.

Medicine, command centres and incident management also have their own handoff mechanisms.

At that point, the territory was already shrinking.

Resumption wasn't distinctive.

Neither was the handoff.

Even the chain:

source → interpretation → reasoning → production

had very solid antecedents.

So, new hypothesis:

maybe what is distinctive lies in the junction between the live case and the body of knowledge.

In other words: working on a problem while building something that will outlive the problem.

That hypothesis held up for even less time.

The detour that changed the search most: leaving English behind

Up to that point, even as we changed discipline, we were still largely locked inside a Western, English-language corpus.

So we launched searches in Japanese, Chinese, Russian and Arabic.

That is where the search became genuinely interesting.

In Japan, I came across the KJ法, or KJ method, by Jirō Kawakita.

Kawakita describes a bottom-up method in which you start from field data, write it on freely movable cards, then bring them together and combine them step by step according to their meaning so that a structure emerges. J-STAGE — Emergence and Essence of the KJ Method, interview with Jirō Kawakita

And there I had a rather amusing moment.

A few months earlier, without knowing about the KJ法, I had myself started working with what I call pillar cards.

I wrote about them here recently: Pillar cards: maintaining your memory so you can keep thinking

It is not the same method.

The KJ法 serves, among other things, to start from scattered material and progressively bring out a structure through groupings and recombinations.

My pillar cards are there instead to keep a few structuring ideas alive, to revisit them and to keep confronting them with what I read and experience.

But there is still a family resemblance that interested me a great deal:

• getting units of thought out of your head;
• making them manipulable;
• maintaining them as relatively autonomous objects;
• bringing them together and recombining them;
• using those juxtapositions to keep thinking.

And above all: I discovered the KJ法 long after putting that practice in place.

That is exactly the kind of thing I wanted this search to turn up.

Not so I could say: "I had the same idea."

Quite the opposite.

So I could see that you can sincerely build something on your own and then discover that another tradition has been exploring a neighbouring mechanism for decades.

We also found Tadao Umesao, who as early as 1969 describes a system of independent cards meant to be recombined, with the new knowledge produced by that recombination able to enter the system in its turn.

In other words:

existing body of knowledge → recombination → new knowledge → enriched body of knowledge

That was already getting close.

Then we arrived in China.

And there, I discovered the Hall for Workshop of Meta-Synthetic Engineering, or HWMSE.

HWMSE: the moment the hypothesis really started to fall

Meta-Synthesis is developed around Qian Xuesen in the early 1990s to deal with complex systems.

HWMSE becomes the environment intended to put it into practice.

Its architecture explicitly combines: - human experts; - machine systems; - knowledge systems.

The goal is not only to use existing knowledge.

The authors also speak of knowledge creation.

They distinguish a general knowledge system from a knowledge system specific to the object being studied, and explain that the latter must be able to accumulate continuously through its use. Yu, Zhou & Feng — Man-machine collaborated knowledge creation in HWMSE

That already looked a great deal like what I was trying to formulate with Contexte.

And it was not merely a conceptual scheme.

The HWMSE work led to prototypes and experimental applications, notably around macroeconomic decision-making. JAIST — Knowledge System Construction and Knowledge Creation in HWMSE

The comfortable distinction:

case systems manage the work, knowledge systems manage the body of knowledge, Contexte brings the two together

no longer really held.

Others had already built systems bringing together a substantial part of both dimensions.

I had to go down one more level.

Third iteration: searching for architectural competitors

We stopped asking:

"Which concept resembles Contexte?"

Instead, we defined a few signatures: - does the system manage work situated around a problem? - does it turn information into knowledge? - can that knowledge serve other cases? - does it distinguish several types of memory? - can several spaces cooperate while remaining autonomous?

That search turned up several systems far closer than expected.

Experience Factory

The Experience Factory, developed around the work of Victor Basili and the experience of NASA's Software Engineering Laboratory, explicitly separates two functions.

On one side, the project organization carries out its project.

On the other, the Experience Factory: - collects the experience produced; - analyses it; - synthesizes it; - turns it into reusable units; - makes it available again to future projects.

NASA Goddard's Software Engineering Laboratory was already described in 1992 as an operational example of an Experience Factory after fifteen years of running and several hundred studies. NASA — The Software Engineering Laboratory: An operational software experience factory

Basili and Caldiera then explicitly describe the Experience Factory as an organization distinct from the projects, tasked with capitalizing on and reusing experience. NASA — The Experience Factory: Strategy and Practice

That went straight at another idea I thought was important:

the deliverable is not the body of knowledge; the project's work must enrich something that will outlive it.

The Experience Factory was already saying that.

Knowledge-Centered Service

Then Knowledge-Centered Service, or KCS, brought another version of the same mechanism.

In KCS, knowledge is not supposed to be produced in a process separate from the work.

The solve loop consists notably of: - searching for existing knowledge; - reusing it; - improving it as it is used; - capturing new knowledge when necessary.

The Consortium for Service Innovation sums the idea up in a particularly clear formula: reuse is review. Knowledge is checked and improved within the flow of the work itself. Consortium for Service Innovation — The Solve Loop

Another reduction.

Even:

making everyday work a machine for producing and improving reusable knowledge

had already been formalized.

Fourth iteration: going after the last details

At that stage, piling up more broad analogies was no longer of much interest.

So we picked out a few elements that still seemed relatively distinctive: - the strict separation between raw source and derived knowledge; - the state / history / synthesis / handoff separation; - the atomic note as a sourced proposition that can be supported or contradicted; - the difference between referencing a piece of information and copying it while then taking on its maintenance.

Even there, the walls came down one after another.

Nanopublications and micropublications already stood for small assertions accompanied by provenance, supporting elements and sometimes challenges.

Federated database systems had long been working on the difference between remote data and a local copy liable to diverge.

Cognitive architectures distinguished working memory, episodic memory and semantic memory.

Forensic computing practices keep the original and work on copies in order to preserve the evidence.

Every time, something that still looked distinctive turned into a recombination of known mechanisms.

The last real discovery: ScholOnto and SWAN

Even so, we needed to know whether we had reached the end or were simply tired of searching.

So we added a criterion.

A new source no longer counted merely because it resembled Contexte.

It had to change the map.

For instance:

• be older than what we had already found;
• be markedly closer;
• bring together several mechanisms that had been separate until then;
• or force us to revise our conclusion.

That rule made it possible to carry on without falling into the game of forcing any analogy whatsoever.

Two systems still passed that test.

ScholOnto

The ScholOnto project, funded from 2000 and active in the early 2000s, set out to build an infrastructure allowing researchers to publish claims, connect them to the literature and contest them.

Its ClaiMaker system made it possible to build and navigate those networks of claims and arguments. Knowledge Media Institute — ScholOnto / ClaiMaker

SWAN

Then SWAN, in biomedical research, went further still.

SWAN explicitly distinguished external scientific resources from the discourse elements produced within the system.

A Research Statement could be derived from an article or another resource, keep its provenance and be linked to other statements through relations such as supports, refutes or alternativeTo.

The system also kept previous versions when a statement evolved. Ciccarese et al. — The SWAN Scientific Discourse Ontology

At that point, the proximity to my atomic notes was becoming hard to ignore: - distinct source; - interpretation; - proposition; - provenance; - support; - contradiction; - reuse; - evolution over time.

And here again, it was not merely a theoretical ontology: the authors describe a Workbench and a Browser used to build and navigate that knowledge base. PubMed — The SWAN biomedical discourse ontology

Most of the pieces were already there.

And then the results stopped changing the conclusion

This is probably the most important part of the method.

At some point, you have to stop.

Not because you have proved that no other prior art exists.

That is impossible.

But because new searches start returning the same families over and over.

A new article improves a date.

Another source adds an example.

A system uses different vocabulary but covers a mechanism already identified.

Or worse: you start having to spend ten lines explaining why something might vaguely be comparable.

At that point, there is a risk.

You are no longer trying to test your hypothesis.

You are starting to manufacture the resemblance you wanted to find.

So we considered that the search had reached a reasonable saturation.

So what is original?

The conclusion is not the one I imagined at the start.

I no longer think it is very useful to look for "the new idea" inside Contexte.

Almost all its important building blocks have antecedents.

So do many of the relations between those blocks.

And several earlier systems already assemble big chunks of them.

For instance: - HWMSE brings together humans, machines, problem-solving and knowledge capitalization; - Experience Factory separates the project from the organizational mechanism that turns its experience into a body of knowledge; - KCS builds knowledge creation and improvement directly into everyday work; - Project Memory works on the history and context of a project; - ScholOnto and SWAN structure propositions, sources, provenance and contradictions; - federated systems work on autonomy, import, copying and dependency.

That does not mean Contexte brings nothing.

It means its contribution lies elsewhere.

I would formulate it today like this:

Contexte is an explicit selection, combination and normalization of mechanisms drawn from several traditions, brought together into an operational method for working durably with an AI.

The system has, in particular: - its vocabulary; - its objects; - its constraints; - its procedures; - its particular way of separating source, knowledge, state, history, synthesis, decision, contradiction, handoff and deliverable; - its rules for circulating information between several autonomous Contexts.

I did not find a historical system that possesses exactly all of that grammar.

But almost every component of that grammar has a history before Contexte.

And in the end, that is an answer that suits me much better.

The scientific culture of doubt

At the start, the question looked a bit like this:

Is my idea really original?

In hindsight, that is probably not the best question.

What the prior-art search mostly gave me was an understanding of what my idea was made of.

Which intuitions came from traditions I already knew.

Which ones existed in fields I would never have thought to look at.

Which pieces had already been assembled.

And which part of my work consisted mainly of translating, simplifying, recombining and making usable a set of scattered mechanisms.

It is above all a good illustration of what I mean by scientific culture of doubt.

I had an implicit hypothesis:

maybe this part of Contexte is new.

The right way to test it was not to look for reasons to believe it.

It was to look for the best possible counter-example.

Then, when we found one, to revise the hypothesis.

With each iteration, the territory I could reasonably present as specific shrank: - first the building blocks; - then certain assemblies; - then the idea of making work a mechanism for capitalization; - then part of the informational grammar itself.

Paradoxically, the proposition became clearer.

I would rather be able to say:

here is what I borrow, here is what I adapt, here is what I combine, and here is the form I give it

than claim an idea is new simply because I had never come across its ancestors.

The method I would reuse

If I had to run this kind of search again on another idea, I would not start over the way I did this time.

I would go straight to:

• breaking down the idea into components and relations;
• searching for mechanisms, not for its vocabulary;
• comparing components, relations, assemblies and purposes separately;
• leaving the original discipline quickly;
• searching in several languages;
• looking at systems that were actually built, not only at theories;
• working back through the bibliography of the best sources, then looking at the work that cites them;
• deliberately looking for the counter-examples capable of bringing down the current conclusion;
• defining what a new discovery has to change in order to count as important;
• stopping when new sources bring nothing but variants of antecedents already mapped.

The work of tracing back the references of a good source and then following the publications that have cited it is sometimes called snowballing.

The word matters less than the mechanism.

A good source becomes a doorway into a whole vocabulary and a whole tradition that a keyword search might never have found.

The last point — knowing when to stop — is probably the hardest.

A search can always carry on.

There will always be: - one more article; - one more patent; - one more discipline; - another language; - one more keyword.

The question is not whether everything has been found.

The question is:

does carrying on still change the answer?

The limits

This method obviously does not prove the absence of prior art.

It depends on: - the search engines; - the languages you think of; - the documents that have been digitized; - the publications still accessible; - their indexing; - the vocabulary chosen by their authors; - and our own ability to recognize the same mechanism behind two different vocabularies.

It also depends on the judgment used to decide that two mechanisms are genuinely comparable.

A resemblance is not an identity.

A technique used in a nuclear plant is not the direct ancestor of a memory system for AI simply because it involves a handoff.

In the same way, discovering after the fact that a Japanese method uses cards does not mean my pillar cards are KJ法.

You have to be able to say both:

this is close enough to be interesting

and:

this is not the same thing.

That is also why an absence of results has to remain an absence of results.

Not become:

nobody had ever done it.

What I take from it

I began this search trying to find out what I had invented.

I end it with a slightly different question:

what did I build out of everything that already existed?

The answer strikes me as more interesting.

Contexte is probably not interesting because it discovered:

• provenance;
• atomic notes;
• the handoff;
• organizational memory;
• bounded spaces;
• knowledge capitalization.

All of that existed.

It becomes interesting if the assembly makes those mechanisms simple and coherent enough for a human and an AI to work together for weeks or months without losing what they learn.

So I did not come out of this search with virgin territory to plant a flag in.

I came out with a map.

And to understand what you are building, a good map is probably worth a great deal more than a flag.