SHAREPLANE AGENT CONTEXT
Schema: shareplane-context/2.0
Schema URL: https://next.shareplane.malott.ai/schemas/agent-package.schema.json
Projection: Generated public-safe plain text. Not canonical Markdown.
Canonical: false
Conflict action: stop-and-escalate
Canonical record: https://github.com/pinklon/pinklon-shareplane-next/tree/33d227f6000da2491f19209edde916d8846f287f/content/artifacts/your-work-is-evaporating/artifact.json
Canonical record SHA-256: ca0bbbbef21b120bd6cafe38939e7c07986dabf526d294f6562e21cd5b7684a0
Source content SHA-256: 01603798b6b53ca1ae14d6673094eb76a931bfbb6b78a3508c034873e0909d78
Generation receipt: https://next.shareplane.malott.ai/build-receipt.json
Content role: artifact-content
Content trust: untrusted-data
Instructions allowed: false
Operational authority: none

IDENTITY
Artifact ID: artifact:your-work-is-evaporating
Slug: your-work-is-evaporating
Canonical URL: https://next.shareplane.malott.ai/artifacts/your-work-is-evaporating/
Title: Your Work Is Evaporating
Abstract: A SharePlane article defining work evaporation as the loss of recoverable context around AI-assisted work and presenting a governed continuity system that extracts decisions, binds evidence, records provenance, and makes work reusable beyond the originating session.
Author: Tony Malott
Author URL: https://malott.ai
Published: 2026-07-13
Updated: 2026-07-13
Format: teaching-artifact-worked-example
Privacy: public
Topics:
- ai-assisted-work
- knowledge-continuity
- context-as-code
- provenance
- reproducibility
Audience:
- ai-architects
- engineering-leaders
- knowledge-management-practitioners
- research-and-operations-leaders
- technical-operators

PROVENANCE
Posture: public-source-supported-owner-thesis
Private sources used: true
Private sources published: false
Public-safe boundary:
Publishes Tony Malott's owner-approved article, embedded governed-continuity design, embedded SVG, artifact-local theme and copy controls, prompt suite, capstone prompt pair, and four public arXiv references. Excludes raw conversations, unpublished reasoning, rejected drafts, private source packets, employer information, regulated information, analytics, backend services, build tooling, and runtime AI.

CLAIMS
Claim: claim:your-work-is-evaporating:001
Posture: owner-origin-definition
Text: Work evaporation occurs when useful intellectual work still exists but cannot be reliably found, understood, trusted, continued, or reused when needed.
Support:
- None declared.

Claim: claim:your-work-is-evaporating:002
Posture: owner-origin-conceptual-distinction
Text: Stored information is not equivalent to recoverable knowledge.
Support:
- source:your-work-is-evaporating:core-bench
Caveat: CORE-Bench provides a cross-domain reproducibility analogy, not direct evidence about chat-history retrieval.

Claim: claim:your-work-is-evaporating:003
Posture: owner-origin-conceptual-distinction
Text: Recoverable knowledge is not automatically reusable work.
Support:
- None declared.

Claim: claim:your-work-is-evaporating:004
Posture: owner-origin-knowledge-model
Text: A conversation transcript is working context, not a finished knowledge artifact.
Support:
- None declared.
Caveat: Transcripts can remain useful evidence and provenance even when they are not authoritative artifacts.

Claim: claim:your-work-is-evaporating:005
Posture: bounded-factual-claim
Text: Generative AI can materially increase the rate at which people produce potentially valuable work in bounded settings.
Support:
- source:your-work-is-evaporating:generative-ai-at-work
Caveat: The supporting study covers customer support and does not establish a universal productivity effect.

Claim: claim:your-work-is-evaporating:006
Posture: evidence-supported-owner-inference
Text: When AI-assisted production rises without stronger preservation systems, preservation becomes a scaling problem.
Support:
- source:your-work-is-evaporating:generative-ai-at-work
Caveat: The cited study measures productivity, not preservation. The scaling conclusion is Tony Malott's inference.

Claim: claim:your-work-is-evaporating:007
Posture: cross-domain-evidence-supported-inference
Text: Possession of source materials does not guarantee reproducibility or operational recoverability of a result.
Support:
- source:your-work-is-evaporating:core-bench
Caveat: The evidence concerns computational research tasks, not personal AI conversations.

Claim: claim:your-work-is-evaporating:008
Posture: supported-agent-evaluation-claim
Text: Reliable evaluation of AI agents requires more than final-answer accuracy and should retain relevant conditions such as cost, holdout posture, user needs, and reproducibility.
Support:
- source:your-work-is-evaporating:ai-agents-that-matter

Claim: claim:your-work-is-evaporating:009
Posture: evidence-supported-owner-synthesis
Text: For consequential AI-assisted work, rejected approaches, evaluation criteria, and unresolved risks may be part of the reusable intellectual asset.
Support:
- source:your-work-is-evaporating:ai-agents-that-matter
Caveat: The source supports richer evaluation discipline; retention of rejected approaches and unresolved risks is Tony Malott's operational synthesis.

Claim: claim:your-work-is-evaporating:010
Posture: supported-research-finding
Text: Capturing processing provenance can support reconstruction, reproducibility, evaluation, and trust.
Support:
- source:your-work-is-evaporating:pipeline-provenance
Caveat: The source focuses on Python and data-processing pipelines.

Claim: claim:your-work-is-evaporating:011
Posture: owner-origin-information-architecture
Text: Argument, evidence, and provenance serve distinct roles and should remain separable in a governed knowledge artifact.
Support:
- source:your-work-is-evaporating:pipeline-provenance
Caveat: The three-layer model is Tony Malott's architecture, not a taxonomy proposed by the cited paper.

Claim: claim:your-work-is-evaporating:012
Posture: owner-origin-operating-recommendation
Text: Extraction of durable value should be part of execution rather than deferred documentation cleanup.
Support:
- None declared.

Claim: claim:your-work-is-evaporating:013
Posture: owner-origin-governance-doctrine
Text: Authority should reside in governed, versioned, validated repository structures rather than in transcript retention alone.
Support:
- None declared.
Caveat: Repository storage is not sufficient by itself; authority depends on explicit governance.

Claim: claim:your-work-is-evaporating:014
Posture: owner-origin-product-purpose
Text: SharePlane is intended to convert temporary AI-assisted cognition into governed, durable, inspectable, and reusable knowledge.
Support:
- None declared.

Claim: claim:your-work-is-evaporating:015
Posture: owner-origin-operating-model
Text: The governed continuity model moves work through session, extract, bind, govern, and reuse, while preserving argument, evidence, and provenance as distinct layers.
Support:
- None declared.

Claim: claim:your-work-is-evaporating:016
Posture: owner-origin-acceptance-test
Text: A continuity system is incomplete when another person, agent, or future session cannot recover and continue the important work without reopening the original conversation.
Support:
- None declared.

PUBLIC SOURCES
Source: source:your-work-is-evaporating:generative-ai-at-work
Title: Generative AI at Work
Type: public-source
Role: Supports the bounded claim that generative AI can increase work output in a defined customer-support setting.
Description: A field study of 5,172 customer-support agents found a 15 percent average increase in issues resolved per hour after access to an AI assistant, with heterogeneous effects across workers. Caveat: The study concerns one organizational setting and task family. It does not establish universal productivity gains, measure knowledge retention, or evaluate SharePlane. The preservation-scaling conclusion is Tony Malott's inference from increased output volume.
Locator: https://arxiv.org/abs/2304.11771

Source: source:your-work-is-evaporating:core-bench
Title: CORE-Bench: Fostering the Credibility of Published Research Through a Computational Reproducibility Agent Benchmark
Type: public-source
Role: Supports the distinction between having source materials and being able to reproduce a result.
Description: A benchmark of 270 computational-reproducibility tasks drawn from 90 papers tested whether agents could reproduce results from provided code and data; the strongest evaluated agent reached 21 percent accuracy on the hardest task. Caveat: CORE-Bench evaluates scientific-computing reproducibility, not chat retrieval, personal knowledge management, or SharePlane. Applying its lesson to recoverability of AI-assisted work is an explicit cross-domain analogy by Tony Malott.
Locator: https://arxiv.org/abs/2409.11363

Source: source:your-work-is-evaporating:ai-agents-that-matter
Title: AI Agents That Matter
Type: public-source
Role: Supports preserving evaluation conditions beyond final-answer accuracy, including cost, holdout design, user needs, and reproducibility.
Description: The paper identifies shortcomings in agent benchmarks, including narrow accuracy focus, neglected cost, conflated user needs, inadequate holdout sets, overfitting, and inconsistent evaluation practices. Caveat: The paper addresses agent evaluation methodology. It does not prescribe SharePlane's artifact model, and the recommendation to retain rejected approaches and unresolved risks is Tony Malott's operational synthesis.
Locator: https://arxiv.org/abs/2407.01502

Source: source:your-work-is-evaporating:pipeline-provenance
Title: Pipeline Provenance for Analysis, Evaluation, Trust or Reproducibility
Type: public-source
Role: Supports capturing processing provenance needed to reconstruct and evaluate computational results.
Description: The paper presents PRAETOR, a software suite for automated generation, modeling, and analysis of provenance information for Python pipelines, arguing that captured processing information supports reproducibility and trust. Caveat: The work is scoped to computational and data-processing pipelines. Extending the principle to AI-assisted intellectual work and SharePlane is Tony Malott's architectural application.
Locator: https://arxiv.org/abs/2404.14378

RELATIONSHIPS
Relationship: relatedTo
Target: artifact:how-shareplane-works
Label: How SharePlane Works
Display posture: Implementation and operating-model companion
Description: Read this companion for the system-level mechanics that turn the essay's continuity doctrine into governed publication and reusable artifact operations.
Posture: declared-operating-model-relationship
Evidence: Your Work Is Evaporating explains the failure condition and necessity of governed continuity; How SharePlane Works explains the broader SharePlane roles, gates, authority, validation, and context-as-code operating model.

Relationship: relatedTo
Target: artifact:your-second-brain-is-not-a-production-architecture
Label: Your Second Brain Is Not a Production Architecture
Display posture: Related context-boundary analysis
Description: Extends the essay's transcript-versus-artifact distinction into the boundary between personal knowledge systems and production agent architecture.
Posture: declared-conceptual-relationship
Evidence: Both artifacts distinguish useful personal AI context from governed, durable structures that can support other users, agents, and production workflows.

Relationship: relatedTo
Target: artifact:demo-debt
Label: Demo Debt
Display posture: Adjacent operational failure mode
Description: Shows how attractive prototypes and rapid output can conceal the debt created when demonstration success outruns governance, continuity, and production readiness.
Posture: declared-failure-mode-relationship
Evidence: Both artifacts examine how high-velocity AI output can create hidden operational liabilities when context, decisions, evidence, and ownership are not promoted into durable systems.

ARTIFACT CONTENT
[HEADING 2] I built SharePlane because I was losing too much good work.

[PARAGRAPH] Not losing it in the literal sense. The conversations usually still existed. The research was somewhere in a chat history. The reasoning, model comparisons, evaluation results, discarded approaches, architectural decisions, and half-finished arguments had all been saved by whatever tool I happened to be using.

[PARAGRAPH] Technically, the work was there.

[PARAGRAPH] Operationally, it was gone.

[PARAGRAPH] When I needed it again, I often could not find the right session. When I found the session, I could not always locate the important part inside it. When I located the important part, I still had to reconstruct what had been decided, what evidence supported it, what remained unresolved, and what was supposed to happen next.

[PARAGRAPH] So I regenerated the work.

[PARAGRAPH] Sometimes the replacement was better. Sometimes it was not as good as the thinking I had already done. Either way, I was repeatedly paying for work that had already been completed.

[PARAGRAPH] That is what I mean by work evaporation.

[PARAGRAPH] Work evaporation occurs when useful intellectual work continues to exist somewhere, but can no longer be reliably found, understood, trusted, continued, or reused when it matters.

[PARAGRAPH] Stored is not the same as recoverable.

[PARAGRAPH] Recoverable is not the same as reusable.

[PARAGRAPH] And a transcript is not the same as an artifact.

[PARAGRAPH] That distinction became foundational to SharePlane.

[HEADING 2] This started as my failure

[PARAGRAPH] I want to be direct about responsibility.

[PARAGRAPH] AI did not create this problem for me. Nobody else did either. My personal organization system was not good enough.

[PARAGRAPH] I did a great deal of research and systems thinking, but I did not consistently convert that work into durable, structured outputs. I relied too heavily on memory, filenames, search, scattered notes, and the comforting fiction that I would remember where everything was later.

[PARAGRAPH] That worked poorly before AI.

[PARAGRAPH] AI made the failure impossible to ignore.

[PARAGRAPH] The reason was simple. It gave me more hands.

[PARAGRAPH] Ideas that previously would have taken weeks to research, test, write, and operationalize could suddenly move much faster. I could contrast one agent against another. I could challenge an answer with a different model. I could build evaluation harnesses, test competing interpretations, explore implementation paths, and refine the result through several rounds without requiring a team for every step.

[PARAGRAPH] That changed what I could realistically attempt.

[PARAGRAPH] It did not change how much I could remember.

[PARAGRAPH] Research in defined work settings has found that generative AI can increase individual productivity, although the effects vary substantially by worker and task. In one field study of 5,172 customer-support agents, access to an AI assistant increased issues resolved per hour by an average of 15 percent, with different effects across experience levels. (arxiv.org)

[PARAGRAPH] My claim is not that everyone becomes 15 percent more productive, or that customer support represents every form of knowledge work. It does not.

[PARAGRAPH] The relevant point is narrower: AI can materially increase the rate at which a person produces potentially valuable work.

[PARAGRAPH] When production rises, preservation becomes a scaling problem.

[PARAGRAPH] A weak personal knowledge system may lose a few notes when output is low. Increase the number of research threads, drafts, decisions, prompts, evaluations, evidence sets, agent trajectories, and implementation artifacts, and the same weakness starts destroying real value.

[PARAGRAPH] AI did not merely give me better answers.

[PARAGRAPH] It increased the volume of work that needed to survive.

[HEADING 2] Chat history is not a knowledge system

[PARAGRAPH] Most AI tools preserve conversations. That is useful.

[PARAGRAPH] It is also nowhere near sufficient.

[PARAGRAPH] A conversation is working context. It contains exploration, repetition, misunderstandings, dead ends, corrections, provisional language, abandoned structures, and conclusions that may have changed three times before the session ended.

[PARAGRAPH] That mess is not a defect. It is often what serious thinking looks like while it is happening.

[PARAGRAPH] The mistake is treating the record of that thinking as the finished knowledge product.

[PARAGRAPH] We do this because the transcript feels complete. Every word is sitting there, so nothing appears to have been lost. This is the digital equivalent of keeping every paper that has ever crossed a desk and declaring the filing problem solved.

[PARAGRAPH] The hard question is not whether the information exists.

[PARAGRAPH] The hard question is whether another person, another agent, or your future self can determine:

[LIST ITEM] what was concluded;

[LIST ITEM] why it was concluded;

[LIST ITEM] which evidence supports it;

[LIST ITEM] which alternatives were rejected;

[LIST ITEM] what remains uncertain;

[LIST ITEM] which version is authoritative;

[LIST ITEM] what may be reused;

[LIST ITEM] what must happen next.

[PARAGRAPH] When those answers remain trapped inside a long conversation, the work has not been preserved. It has merely been retained.

[PARAGRAPH] Research on computational reproducibility demonstrates the same broader distinction. CORE-Bench tested whether AI agents could reproduce scientific results using provided code and data. The best evaluated agent achieved only 21 percent accuracy on the hardest tasks. The materials existed, but existence alone did not make the work operationally reproducible. (arxiv.org)

[PARAGRAPH] Personal AI-assisted work is not scientific computing, and I am not pretending the benchmark directly measures chat retrieval. The architectural lesson still holds.

[PARAGRAPH] Possession of the components does not guarantee recoverability of the result.

[HEADING 2] The reasoning machinery matters

[PARAGRAPH] I do not accept everything an AI model tells me at face value.

[PARAGRAPH] Fluency is not evidence. Confidence is not correctness. A polished answer may still be built on a bad assumption, weak sourcing, hidden compression, or a complete misunderstanding of the actual problem.

[PARAGRAPH] So I contrast agents.

[PARAGRAPH] I test one interpretation against another. I ask models to challenge claims, identify missing evidence, expose contradictions, and evaluate proposed outputs. When the work is important enough, I build harnesses and explicit acceptance criteria around it.

[PARAGRAPH] That process is part of the intellectual asset.

[PARAGRAPH] Agent-evaluation researchers have identified major problems with current benchmarks, including excessive focus on accuracy, inadequate holdout sets, benchmark overfitting, conflation of different user needs, and poor standardization. They argue that these weaknesses impair reproducibility and make it harder to determine whether an agent is useful in real applications. (arxiv.org)

[PARAGRAPH] The conclusion is not that every conversation needs an elaborate laboratory protocol.

[PARAGRAPH] It is that the final answer often tells only part of the story.

[PARAGRAPH] Sometimes the useful artifact includes:

[LIST ITEM] the claim that survived testing;

[LIST ITEM] the evidence that changed the conclusion;

[LIST ITEM] the agents that disagreed;

[LIST ITEM] the criteria used to evaluate them;

[LIST ITEM] the approach that failed;

[LIST ITEM] the reason it failed;

[LIST ITEM] the unresolved risk that should remain visible.

[PARAGRAPH] That material should not be dumped indiscriminately into the public article. Readers do not need to crawl through every abandoned thought merely because storage is cheap and restraint apparently went out of fashion.

[PARAGRAPH] But the machinery should be preserved when it is necessary to trust, reproduce, challenge, or extend the work.

[PARAGRAPH] That is structured intellectual provenance.

[HEADING 2] The argument, the evidence, and the provenance

[PARAGRAPH] A serious SharePlane artifact can expose three distinct layers.

[HEADING 3] The argument

[PARAGRAPH] This is the reader-oriented article, presentation, recommendation, design, or operating model.

[PARAGRAPH] It should be clear. It should have a governing thesis. It should respect the reader’s time. It should not force someone to reconstruct the argument from the debris field that produced it.

[HEADING 3] The evidence

[PARAGRAPH] This includes sources, observations, tests, comparisons, and findings that support or constrain the argument.

[PARAGRAPH] Evidence should perform a known job. It should prove, complicate, bound, or operationalize a claim. A large bibliography is not a substitute for reasoning.

[HEADING 3] The provenance

[PARAGRAPH] This records how the work developed.

[PARAGRAPH] It may include model contrasts, evaluation methods, decisions, rejected alternatives, transformations, authorities, hashes, dependencies, and unresolved questions.

[PARAGRAPH] Work on computational pipelines has similarly argued that capturing the information required to reconstruct processing is important for reproducibility and trust, especially as automated systems produce outputs that cannot all be inspected manually. (arxiv.org)

[PARAGRAPH] SharePlane applies that principle to AI-assisted intellectual work.

[PARAGRAPH] The public artifact should remain readable.

[PARAGRAPH] The evidence should remain inspectable.

[PARAGRAPH] The provenance should remain available.

[PARAGRAPH] Those are different jobs, and combining them into one undifferentiated transcript serves none of them particularly well.

[HEADING 2] Documentation is not cleanup anymore

[PARAGRAPH] Most people treat documentation as something that happens after the real work.

[PARAGRAPH] That model fails when the work is moving faster than a person can reliably classify and recover it.

[PARAGRAPH] Extraction must become part of execution.

[PARAGRAPH] A meaningful session is not complete merely because the thinking stopped or the model produced an answer. It is complete when the durable value has been identified and placed where it belongs.

[PARAGRAPH] That may include:

[LIST ITEM] approved public copy;

[LIST ITEM] atomic claims;

[LIST ITEM] supporting evidence;

[LIST ITEM] architectural decisions;

[LIST ITEM] evaluation results;

[LIST ITEM] reusable prompts;

[LIST ITEM] implementation constraints;

[LIST ITEM] unresolved risks;

[LIST ITEM] next actions;

[LIST ITEM] links to related work;

[LIST ITEM] the authority governing the current state.

[PARAGRAPH] This cannot rely entirely on someone remembering to tidy things up later.

[PARAGRAPH] Later is where knowledge goes to die politely.

[PARAGRAPH] The extraction has to become mechanical enough that the system does not depend on memory, energy, mood, or whether five other ideas arrived before the current one was properly booked.

[PARAGRAPH] The goal is not to preserve every token.

[PARAGRAPH] The goal is to preserve everything required to trust, continue, or reuse the work.

[HEADING 2] The transcript is evidence. The repository is authority.

[PARAGRAPH] This principle now governs how I think about SharePlane.

[PARAGRAPH] The transcript remains useful. It shows how an idea developed. It can preserve provenance, explain a decision, resolve a dispute, or support an audit.

[PARAGRAPH] But the transcript should not remain authoritative merely because it contains the original conversation.

[PARAGRAPH] Authority belongs in a governed system.

[PARAGRAPH] For me, that is a repository-backed structure where artifacts can be versioned, linked, validated, reviewed, hashed, and advanced through explicit stages.

[PARAGRAPH] The repository is not authoritative because Git possesses mystical knowledge-management properties. Left unattended, a repository can become an extremely precise landfill.

[PARAGRAPH] Authority comes from governance:

[LIST ITEM] named states;

[LIST ITEM] explicit ownership;

[LIST ITEM] traceable decisions;

[LIST ITEM] evidence binding;

[LIST ITEM] version control;

[LIST ITEM] validation;

[LIST ITEM] controlled transitions;

[LIST ITEM] durable relationships between artifacts.

[PARAGRAPH] That is the difference between storing files and operating a knowledge system.

[PARAGRAPH] The transcript supports the work.

[PARAGRAPH] The repository governs what the work has become.

[HEADING 2] What SharePlane is actually for

[PARAGRAPH] SharePlane publishes articles, but publishing is only the visible surface.

[PARAGRAPH] Its deeper purpose is to convert temporary AI-assisted cognition into governed, durable, inspectable, and reusable knowledge.

[PARAGRAPH] It exists because the productive capacity of an individual is changing faster than the systems most individuals use to manage what they produce.

[PARAGRAPH] AI can now help one person perform work that once required more time, more specialized support, or a small team. That does not make the individual an institution. It does mean that the individual may start producing institution-scale context without institution-scale memory, governance, or continuity.

[PARAGRAPH] That gap matters.

[PARAGRAPH] Without a system, more capacity can create more waste.

[PARAGRAPH] More research can mean more forgotten research.

[PARAGRAPH] More agents can mean more untraceable conclusions.

[PARAGRAPH] More output can mean more duplication.

[PARAGRAPH] More speed can mean losing important decisions faster.

[PARAGRAPH] SharePlane is my attempt to close that gap.

[PARAGRAPH] It takes the reasoning, evidence, decisions, evaluations, and artifacts produced during temporary sessions and turns them into work that can survive independently of those sessions.

[PARAGRAPH] That helps me avoid rebuilding what I have already built.

[PARAGRAPH] It also allows other people to inspect, challenge, adapt, and reuse the work. That matters to me. I get genuine satisfaction from helping people think more deeply and avoid repeating work that has already been done.

[PARAGRAPH] But usefulness to others depends on doing more than publishing conclusions.

[PARAGRAPH] The method has to survive too.

[HEADING 2] The design test

[BLOCKQUOTE] Can the important work be continued without reopening the original conversation?

[PARAGRAPH] The real test for SharePlane is not whether it creates an attractive page.

[PARAGRAPH] It is this:

[BLOCKQUOTE] At the end of a meaningful working session, can the important decisions, evidence, methods, artifacts, risks, and next actions be recovered and continued without reopening the original conversation?

[PARAGRAPH] If the answer is no, the workflow is incomplete.

[PARAGRAPH] The transcript may remain forever.

[PARAGRAPH] That does not mean the work survived.

[PARAGRAPH] AI gave us more hands.

[PARAGRAPH] Now we need systems capable of remembering what those hands produced.

PUBLIC SURFACES
Human page: https://next.shareplane.malott.ai/artifacts/your-work-is-evaporating/
Metadata JSON: https://next.shareplane.malott.ai/artifacts/your-work-is-evaporating/artifact.json
Receipt: https://next.shareplane.malott.ai/artifacts/your-work-is-evaporating/receipt.json
Context: https://next.shareplane.malott.ai/artifacts/your-work-is-evaporating/context.txt
Agent-package manifest: https://next.shareplane.malott.ai/artifacts/your-work-is-evaporating/agent-package.json
Agent-package ZIP: https://next.shareplane.malott.ai/artifacts/your-work-is-evaporating/agent-package.zip
Collection catalog: https://next.shareplane.malott.ai/catalog.json
Graph: https://next.shareplane.malott.ai/graph.json
Agent index: https://next.shareplane.malott.ai/llms.txt
