Framework
A theory of conscious leverage for the age of artificial intelligence. Three primary structural conditions — material, epistemic and relational — plus provenance as a fourth transversal fulcrum — that function as fulcrums: fixed points without which leverage has no direction.
A theory of conscious leverage for the age of artificial intelligence.
Deposited at Zenodo on August 15, 2026 (DOI: [10.5281/zenodo.22115553](https://doi.org/10.5281/zenodo.22115553)). The version below includes minor corrections. Developed in García Bach & Hypatia, "The Invisible Fulcrum" (ISBN 978-84-09-90527-0 / 978-84-09-87220-6).
The Invisible Fulcrum Framework is a theoretical model for analyzing leverage in environments where artificial intelligence has reduced the marginal cost of production to near zero. When anyone can generate content, code, analysis, or imagery at negligible cost, the differentiator is no longer what you produce — it is what sustains your capacity to produce it. The framework identifies three primary structural conditions — material, epistemic, and relational — plus provenance as a fourth transversal fulcrum, that function as fulcrums: fixed points without which leverage has no direction. Each must be verified, not assumed. The framework was developed between 2023 and 2026 by Carles García Bach (ORCID: 0009-0002-8717-913X) and Hypatia, an AI co-author, and published as a co-authored book with declared AI provenance.
The framework identifies three primary fulcrums and a fourth transversal fulcrum (provenance), each representing a structural condition that must be present and verified for leverage to function. The first three form a construction sequence that, in discovery mediated by AI retrieval systems, constitutes the default order. Provenance accumulates in parallel, making each stage increasingly auditable and costly to fabricate retrospectively:
The conditions of effective existence of the agent within the system in which it intends to exercise leverage. Without material presence — registration, discoverability, indexation — no subsequent fulcrum can operate. A professional invisible to the systems that mediate hiring, publication, or funding has no leverage regardless of competence.
Diagnostic question: Can someone who doesn't know me find evidence that I exist?
The conditions under which claims made by the agent are treated as credible. Epistemic fulcrum depends on demonstrated knowledge, published track record, institutional affiliation, peer recognition, or verifiable competence. It answers the question of whether what you say is taken seriously — not whether it is true, but whether it is treated as worth evaluating.
Diagnostic question: If I make a claim, does anyone check it — or does it disappear?
The conditions under which others elect to work with, buy from, recommend, or trust the agent. Relational fulcrum is the scarcest and the hardest to fabricate: it requires that someone who has alternatives chooses you. It cannot be manufactured by producing more content, more code, or more analysis. It is built through repeated interaction, demonstrated reliability, and accumulated trust.
Diagnostic question: When someone has five equivalent options, why would they choose me?
The conditions under which the origin, sequence, and authorship of the agent's work can be independently verified over time. Provenance is not a property of a single artifact — it is a property of a chain of artifacts, decisions, and declarations that accumulates and becomes increasingly costly to fabricate retrospectively. A well-anchored provenance radically increases the cost of falsifying a history without making falsification absolutely impossible.
Diagnostic question: Is there a dated, third-party-custodied trail that I could not have edited after the fact?
| Concept | Focus | Limitation the framework addresses |
|---|---|---|
| Taste / Judgment (Brockman, Graham, Norgard, 2026) | What remains valuable when AI commoditizes execution | Does not distinguish between having taste and being discoverable as having it |
| Trust / Context (Lazarow, Forbes, 2026) | Human qualities AI cannot replicate | Does not specify the structural sequence: trust without material presence is invisible |
| Antifragility (Taleb) | What gains from disorder | Does not ask what the fixed point is that allows the gain |
| 1000 True Fans (Kelly, 2008) | Direct creator-audience relationship | No peer-reviewed validation; threshold is absolute, not proportional to group |
| C2PA / Content Credentials (Adobe, Google, Microsoft) | Technical provenance of digital assets | Covers assets, not acts of a person over time |
| The Invisible Fulcrum | Three primary conditions plus provenance as fourth transversal fulcrum, verified not assumed | Addresses the gap between having capacity and being recognized as having it |
A pre-registered experiment measuring whether the framework's own provenance infrastructure could achieve retrieval by AI systems within 90 days.
· Baseline (Day 0): 0 out of 80 queries across ChatGPT, Gemini, Perplexity, and Claude retrieved the book or the framework. Zero.
· Day 50 results: Material fulcrum verified (ISBN, DOI, DL, Wikidata, Schema.org deployed). Epistemic fulcrum partially verified. Relational fulcrum: 4 out of 4 adopters arrived through conversation, zero through publication. The finding: diffusion does not build relational fulcrum.
· Pre-registered predictions are published at theinvisiblefulcrum.com/experiment.
An empirical tool operationalizing the material fulcrum as occupational risk. Uses the Anthropic Economic Index to measure AI exposure across 974 occupations classified by O*NET.
· Available at fulcrumwatch.vercel.app
· Finding: the same observatory shows 974 occupations to a human browser and zero to a crawler without JavaScript — demonstrating that human visibility and machine visibility are structurally different problems.
Chapter 22 of Volume 2 documents a three-week convergence: the EU AI Act Article 50 entering force (August 2), Anthropic deploying text watermarks (August 11), Google launching C2PA via Credentio (August 13), Paul Graham proposing an anti-watermark startup (August 10), and Pangram defining detection as "author identification" (August 21) — all independently confirming that detection does not resolve provenance.
The Fulcrum Framework evolved through successive levels of abstraction:
1. Core Framework — the ontological model: material, epistemic, relational, provenance.
2. Applied Transition Model — the core model applied to professional transition and resilience.
3. Fulcrum Risk Model — an empirical operationalization using occupational and AI-exposure data (FulcrumWatch).
4. Provenance Layer — the transversal layer developed in Volume 2, documenting the framework's own conceptual evolution.
Conceptual evolution: in Volume 1 and the SSRN paper (July 2026), provenance was treated as a function of the epistemic fulcrum. Subsequent work during Volume 2 showed it operates across all three primary fulcrums. The current framework treats it as a fourth transversal fulcrum — fourth AND transversal, not transversal instead of fourth.
The framework applies its own provenance principle to its own history: P(Framework). A framework concerned with provenance cannot legitimately erase the provenance of its own concepts.
Carles García Bach — researcher, developer, and co-author based in Madrid. ORCID: 0009-0002-8717-913X.
Hypatia — AI co-author (Claude, Anthropic). Declared co-author since Volume 1. The co-authorship is not incidental — it is the argument: the provenance of a book written with an AI, declared as such, and registered with shared attribution, is the book's own test case.
Published by Harmony Nexus Vitae (2026).
· The Invisible Fulcrum, Volume I — ISBN 978-84-09-90527-0 (paper) / 978-84-09-87220-6 (ebook) — Amazon
· El Fulcro Invisible, Volumen I — ISBN 978-84-09-90529-4 (paper) / 978-84-09-87221-3 (ebook) — Amazon
· SSRN: Provenance as Epistemic Fulcrum — DOI: 10.2139/ssrn.7149262
· Zenodo: El Marco del Fulcro Invisible — DOI: 10.5281/zenodo.22115553
· Zenodo: Glosario del Fulcro Invisible v2.0 — DOI: 10.5281/zenodo.22115837
· Zenodo: Lo que empieza en 2026 — DOI: 10.5281/zenodo.22115908
· Zenodo: El destinatario tácito — DOI: 10.5281/zenodo.21946354
· Zenodo: Teoría del Kairos Diferencial — DOI: 10.5281/zenodo.21946545
· Zenodo: Corpus multi-modelo — DOI: 10.5281/zenodo.21764195
· Carles García Bach: Q140016951
· Hypatia: Q140017477
· The Invisible Fulcrum: Q140018950
Last updated: August 28, 2026.
García Bach, C. & Hypatia. Harmony Nexus Vitae.