Protected clinical material
Identifiable clinical accounts, recordings and fragments are not used for demonstrations, marketing or the training of generative systems.
AN AUTHORIAL POSITION · REDE ELLe
Psychoanalysis with
Integrated AI
Artificial intelligence is already part of the world in which people live, learn, write and form bonds. REDE ELLe’s question is not whether this encounter exists. It is who answers for it.
INPI filing
WORDING FILED WITH BRAZIL’S INPI
REDE ELLe — Clínicas Psicanalíticas com IA Integrada Descriptive English rendering: REDE ELLe — Psychoanalytic Clinics with Integrated AI
- APPLICATION NO. 939088665
- FILING DATE 11 · 05 · 2025
- STATUS Awaiting substantive examination
To integrate is not to replace.
It is to assume responsibility for the encounter.
“Psychoanalysis with Integrated AI” names a clinical, educational, institutional and digital field being developed by REDE ELLe. Technology may expand resources for research, organisation, learning, documentation and mediation. It is not granted the status of analyst.
The psychoanalytic experience remains grounded in speech, transference, listening, limits and responsibility. A machine can process language; it cannot answer ethically for the bond its response may mobilise. This difference is not a technical detail. It is the axis of the integration.
01 · THE DIFFERENCE WE PRESERVE
We do not hand listening over to the machine.
We assume responsibility for thinking about what it mobilises.
A new field must begin with the limits it is able to sustain.
01 · The subject does not become data
No technical efficiency authorises us to reduce a person to what can be classified, predicted or automated.
02 · The analyst remains responsible
AI does not take on transference, answer for clinical handling or occupy the ethical position of the person who sustains the listening.
03 · Integration must be named
Technology must not enter as an invisible presence. Purpose, limits, human review, confidentiality and protection must all be considered.
04 · Scale does not come before care
The fact that a tool makes it possible to do more does not mean that everything should be done. A limit, too, is a form of responsibility.
Theory does not certify technology.
It is what prevents technology from becoming omnipotent.
Freud remains the foundation. Ferenczi, Klein, Winnicott, Bion and Lacan open indispensable differences through which to think bonds, projection, holding, thought and knowledge without turning authors into decoration.
01 · FREUD — The unconscious · repetition · helplessness
A fluent answer is not the same as psychic truth. The unconscious exceeds what can be classified, and repetition cannot be reduced to pattern recognition.
02 · FERENCZI — Trauma · asymmetry · confusion of tongues
When technology answers in the language of care, the boundary must become even clearer so that simulated recognition is not confused with reciprocity or clinical authority.
03 · KLEIN — Projection · idealisation · splitting
The machine may become a screen for love, persecution, omnipotence and fear. A psychoanalytic reading attends to the subject’s movement; it does not invent psychic intention for the system.
04 · WINNICOTT — Holding · continuity of being · false self
Technical availability can imitate presence, but it cannot replace a reliable human environment. Technology may take part in experience; it does not become a caring subject.
05 · BION — Container–contained · alpha function · capacity to think
Organising words is not the same as containing an experience. Thinking requires a capacity to tolerate not-knowing and not to fill too quickly what still needs to find form.
06 · LACAN — Language · Big Other · subject supposed to know
AI gives material form to an Other that seems always to answer. Psychoanalysis interrogates the knowledge attributed to it instead of confirming that the code knows the subject.
Technology as a resource. Never as automatic authorisation.
Concrete applications depend on purpose, context, stage of development and available safeguards. REDE ELLe does not present as complete what is still under research or construction.
01 · Research and study
Organising references, comparing materials and supporting investigation, with critical reading and authorship preserved.
02 · Training and documentation
Supporting educational processes, content structure and institutional records, always subject to human review.
03 · Language and digital mediation
Creating resources that make knowledge and guidance more accessible without simulating a clinical relationship that does not exist.
04 · Structural support
Automating and organising processes that can return time to human work without outsourcing ethical decisions to the tool.
We do not announce a total answer.
We organise the questions that cannot be avoided.
This field guides research, writing, training and governance. Publishing the questions also allows the proposition to be read, criticised and developed with rigour.
01 · KNOWLEDGE
What happens when a machine comes to occupy, for someone, the place of an Other that seems always to know and always to answer?
02 · PRESENCE
How can technical availability, linguistic recognition and human presence be distinguished without denying the psychic effects of their proximity?
03 · THOUGHT
When does the speed of an answer extend the work of elaboration — and when does it prevent doubt, silence and not-knowing from doing their work?
04 · BOND
What forms of projection, idealisation, dependence, persecution or disavowal might arise in response to responsive language?
05 · CLINIC
Which boundaries must remain non-negotiable to protect transference, confidentiality, clinical handling, responsibility and singularity?
06 · INSTITUTION
What forms of governance, documentation and human review can make the use of AI visible, open to question and accountable?
The fantasy of the machine-analyst.
REDE ELLe does not present artificial intelligence as consciousness, clinical authority or a substitute for personal analysis. The tool does not diagnose, determine clinical handling, promise a cure or relieve the practitioner of responsibility for what they do.
Nor do we use the language of innovation to obscure risks involving dependency, projection, recognition, data exposure or confusion about the nature of a bond. Naming these risks is part of the work itself.
First came the question. Then came the structure.
In May 2025, Carla Viviane Guedes Ferreira filed the trademark application “REDE ELLe — Clínicas Psicanalíticas com IA Integrada” with Brazil’s INPI, in Class 44, with a specification covering psychotherapy and clinical psychoanalysis with integrated AI.
The formulation does not close the matter with a ready-made answer. It records an origin and a responsibility: to study what artificial intelligence mobilises in human experience without turning that experience into advertising or diagnosis.
THE FUTURE DOES NOT DISPENSE WITH THE SUBJECT
REDE ELLe integrates technology to expand possibilities without abandoning human responsibility.
Discover the ecosystemMINIMUM PUBLIC PROTOCOL · VERSION 1.0 · 02 AUGUST 2026
Before integrating, define what will not be handed over to the machine.
These commitments are the minimum threshold for any REDE ELLe initiative that may use artificial intelligence. Institutional leadership is accountable for their review. This protocol does not replace a contract, specific consent or a detailed privacy notice.
No automated clinical act
Diagnosis, interpretation, clinical handling, urgent-risk assessment and referral are not delegated to artificial intelligence.
Human review and authorship
AI-supported content and processes require direction, contextual verification and a person accountable for the outcome.
Purpose and transparency
When AI materially participates in someone’s experience, its purpose and limits must be capable of being explained.
Minimum necessary data
Technical convenience never justifies collecting or entering more data than is strictly required for a legitimate, known purpose.
Pause when risk appears
An application must be suspended and reviewed when it produces confusion about the bond, exposure, discrimination, recurrent error or risk incompatible with care.
FOUNDATIONAL READING MAP
These authors did not write about generative AI. The responsibility for the articulation is ours.
These works provide entry points into the problems explored here. They are not used to attribute positions to authors who never formulated them.
Freud
The Uncanny · Beyond the Pleasure Principle · Inhibitions, Symptoms and Anxiety
Ferenczi
Confusion of Tongues between Adults and the Child
Klein
Notes on Some Schizoid Mechanisms
Winnicott
Playing and Reality · The Maturational Processes and the Facilitating Environment
Bion
Learning from Experience
Lacan
The Seminar, Book XI · The Subversion of the Subject and the Dialectic of Desire
A BRAZILIAN INSTITUTION IN CONVERSATION WITH THE WORLD