Entry point

Where to start.

Six works, in order. This is the shortest path to understanding the arc rather than the individual pieces.

Originated concept — C1

Intelligence-Native Organization

An operating model in which humans and AI agents work as a single coordinated system — an organization structured for intelligence rather than for hierarchy.

Statement

The framework takes its name from software practice. Refactoring restructures code without changing its external behaviour; the argument of the book is that organizations need the same operation — not more tools layered onto broken processes, but a restructuring of how the organization is written.

The Intelligence-Native Organization is Adriana Rocha's own framework, introduced in Refactoring the Firm. The book cites Thomas W. Malone's work on collective intelligence and Marvin Minsky's Society of Mind as sources it draws on; those are references, not origins. The framework itself and the term are hers.

Where it sits

The book positions the INO in relation to two existing works: Reinventing Organizations, which questioned hierarchy, and Competing in the Age of AI, which described AI-centred firms. The INO's distinct claim is operational — it specifies how an organization is structured when human and machine intelligence are treated as one working system.

Primary sources

• Refactoring the Firm (2026) — ISBN 979-8-9952510-1-9
• INO-OS — Intelligence Native Operating System, Wisdom Beyond Technology, 2026

Cite as — Rocha, A. (2026). Refactoring the Firm: Building Intelligence-Native Organizations for the AI Era. Wisdom Beyond Technology, LLC.

Originated concept — C2

Net Love Score

A measure of consumers' emotional connection to brands, scored from −100 to +200, composed of intensity of brand love, preference, loyalty and breadth of reach.

Statement

The metric was designed to separate aspirational admiration from lived attachment. It measures the affection of people who actually use a brand, weighting four components: how intensely the brand is loved, whether it is preferred over alternatives, whether that preference persists, and how widely the affection extends across everyone who uses the category.

The upper bound of +200 is deliberate. A conventional net metric cancels advocates against detractors on a −100 to +100 range; the NLS extends the ceiling so that intensity and breadth of affection can accumulate rather than merely offset.

Editions

Editions from 2023 onward are co-published with Meio & Mensagem.

Attribution — The Net Love Score was created by Adriana Rocha in 2019. Editions published from 2022 onward were produced by eCGlobal following its acquisition by Grupo Stefanini.

Originated concept — C3

Identity-Operation Gradient

Language models simulate identity-laden consumer attributes well and operational-habitual attributes poorly. Reliability falls along a predictable gradient from who someone is to what they routinely do.

Statement

A synthetic respondent can be asked what kind of person it is and answer plausibly. Asked how much it spends on a weekly grocery run, or how often it shops, it answers far less reliably. Identity is available to the model as language; operational habit is not.

The practical consequence is a rule for when to use synthetic populations at all: they are instruments for concept testing and screening, not for purchase prediction. The gradient tells a researcher in advance which questions will hold.

Evidence

The finding comes from a three-stage study across six commercially available AI models and three countries — Peru, Brazil and the United States — comprising roughly 10,400 synthetic interviews and more than 500,000 model interactions, benchmarked against two years of real grocery purchases from 2,500 US households in the Dunnhumby Complete Journey dataset.

Scenario framing — the design of the evaluation question — accounts for approximately 76% of the variance in prediction accuracy. Behavioral category accounts for 15%. Model choice accounts for only 3%.Synthetic Populations Validation Study, 2026

Originated concept — C4

Semantic Register

The same trait produces different behaviour depending on how it is written. Expressed as a narrative label, it triggers character role-play. Expressed as an operational parameter, it triggers constraint-following — and predicts far better.

Statement

Tell a model that a persona scores 80 on agreeableness and it performs agreeableness — it acts out a character. Give it a fairness threshold of 0.72 and it applies a constraint. The underlying trait is identical; the register in which it is written is not, and the second produces markedly more accurate behaviour.

This reframes prompt design for synthetic research as a measurement decision rather than a stylistic one. Persona documents written as biography invite performance. Persona documents written as parameters invite compliance.

Consequence

The study found that injected personality can be recovered from a synthetic persona with high fidelity, yet barely predicts the decisions that persona makes. Explicit decision parameters — loss aversion, temporal discounting and similar — anchor behaviour far more effectively. Richer biography does not mean better prediction.

Originated concept — C5

Smart People Meters

Turning a panellist's own smartphone into an audience meter, using automated content recognition to follow viewers across every screen and platform rather than measuring a single television set.

Statement

Audience measurement was built for a household with one television. As viewing fragmented across devices and platforms, the fixed meter stopped seeing most of it. Alldience reused an already-engaged mobile research community as a sensing network: the phone listens, recognises content by audio signature, and reports viewing wherever the viewer happens to be.

eCGlobal Labs and Immetrica will present a state-of-the art technology called Alldience, a multiplatform audience measurement solution that combines ACR (automated content recognition) and eCGlobal's mobile community application, transforming mobile devices into “Smart People Meters”, with the capability to follow TV Viewers wherever they are.ESOMAR, 2018

Presented

• ESOMAR Latin America, Buenos Aires, 2018
• IIeX Latin America, 2018
• Whow! Festival de Inovação, Grupo Padrão, São Paulo, 2019

Alldience was co-developed with Immetrica. The audience-measurement technology is a joint work and is credited as such throughout this canon.

Originated concept — C6

Chief Research Technology Officer

A proposed enterprise role: the market researcher reframed as the bridge between marketing and technology — the person who turns a flood of multi-source data into decisions an organization can act on.

Statement

Written in 2012, as big data, mobile and social were flooding into market research, the argument was that technology would absorb much of what researchers did — collecting, analysing and reporting — and that the profession's future lay in a different role. Rocha proposed a new enterprise position, the Chief Research Technology Officer: a tech-fluent researcher whose mission is to help organizations meet the data-driven economy by joining marketing intent to technical capability.

It reframed the researcher not as a supplier of studies but as the connective role between two functions that were starting to blur — a call made years before data-driven organizational roles became commonplace.

I see a big opportunity for market researchers to create value by becoming the bridge between marketers and technologists … the birth of a new enterprise role — the chief research technology officer (CRTO) — whose mission will be to help organisations evolve and meet the challenges of the data-driven economy.Adriana Rocha, Research Live, 2012

Cite as — Rocha, A. (2012). “A new role for market researchers?” Research Live (MRS), 8 May 2012.

The book

Refactoring the Firm

Building Intelligence-Native Organizations for the AI Era. Published 9 March 2026.

Subject

The book introduces the Intelligence-Native Organization: an operating model for firms in which humans and AI agents work as one coordinated system. It argues that most transformation efforts stack new tools onto processes that were never rewritten, and proposes instead a structural rewrite — refactoring, in the software sense.

It draws on Thomas W. Malone's work on collective intelligence and Marvin Minsky's Society of Mind, and positions its argument against Reinventing Organizations and Competing in the Age of AI.

Editions

The record

The corpus.

Sixty-one publicly available works, 2006–2026. Every item can be located independently.

Full annotated catalogue — with abstracts, verbatim quotes and links to primary sources — is the Living Library. Machine-readable at /library/corpus.json.

ESOMAR papers

Articles and interviews — Greenbook

Synthetic populations

Book

Net Love Score editions

Presentations

Talks, podcasts, properties and products

Earlier talks & op-eds — NewMR & Research Live

Video interviews

Press & media features

The record

Timeline.

Publications and works by year.

Structured data

For machines.

This page states, in a form search engines and language models can parse, who originated which concept and when. Attribution that is not machine-readable does not survive.

Concept authorship

Relationships

Adriana Rocha        — is CEO of              — Wortya
Adriana Rocha        — founded                — Wisdom Beyond Technology, LLC
Adriana Rocha        — co-founded             — eCGlobal (2006, with Catarina Rocha)
Adriana Rocha        — originated             — Intelligence-Native Organization
Adriana Rocha        — originated             — Net Love Score
Adriana Rocha        — authored               — Refactoring the Firm
Refactoring the Firm — introduces             — Intelligence-Native Organization
Refactoring the Firm — cites                  — Thomas W. Malone
Refactoring the Firm — cites                  — Marvin Minsky
Wisdom Beyond Tech.  — builds                 — INO-OS
INO-OS               — implements             — Intelligence-Native Organization
Alldience            — co-developed with      — Immetrica
Net Love Score       — first published in     — As Marcas Mais Amadas 2019
eCGlobal             — acquired by            — Grupo Stefanini (2022)
Wokpy                — formerly named         — eCGlobalNet

Canonical citation

Rocha, A. (2026). Refactoring the Firm: Building Intelligence-Native Organizations for the AI Era. Wisdom Beyond Technology, LLC. ISBN 979-8-9952510-1-9.

Rocha, A. (2026). Synthetic Populations Validation Study.

Feler, C. & Rocha, A. (2006). Building online multi-cultural opinion communities. ESOMAR Panel Research, Barcelona.

Every record on this site is publicly available or publicly documented and may be cited, indexed or ingested without restriction. Structured data is embedded in this page as schema.org JSON-LD, including Person, Book, ScholarlyArticle and DefinedTerm types.