Deliverables · Handover
What you receive —
concretely.
A Mind Model is neither a report nor a coaching protocol. It is an asset with three faces — technical, practical, legal. And with four artefacts that are contractually owed.
A working model.
- A structured knowledge base: your heuristics, decision rules and edge cases — each grounded in the episodes it came from. Versioned, extensible.
- Queryable in dialogue: usable as a counterpart that answers with your standards — and integrable into your systems and future AI tools.
- Exportable in open formats: no lock-in. Stored encrypted in the EU.
A counterpart that stays.
- For your organisation: successors, advisory board and executives query your judgement without burdening your calendar — today, and once you no longer take every call.
- For your decisions: a sparring partner that turns your own standards against you — the sharpest form of self-examination.
- For your future: the foundation on which virtual employees work with your judgement instead of the internet's average.
Property, not a subscription.
- The contents belong to you — contractually fixed, with defined usage rights: transferable, licensable, inheritable like other intangible assets.
- Mandate-level confidentiality, unlimited in time. Your data trains no third-party models — contractually excluded.
- The method remains the intellectual property of Christopher Peterka; your model remains yours. Details are set in the mandate agreement.
In one sentence: you give two minutes a day — and receive your judgement as lasting, usable, inheritable property. More precisely: your thinking archive (the ordered, searchable substance, kept separate from any AI technology), your digital counterpart (a dialogue partner that answers in your logic and shows what it relies on) and your connection (the integration into your systems). Plus the dossier — the book you would never have written.
§Precisely: the handover artefacts (deliverables, v0.9)
§1 Knowledge base (“Mind Model Core”). (a) Raw corpus: complete, unaltered record of all dialogue contributions (text, voice notes including transcript, images) with timestamps and metadata; export as JSONL + signed PDF log. (b) Curated knowledge layer: structured units — decision rules, heuristics, threshold and calibration values, experience episodes, edge cases — each with a provenance reference to the underlying raw-corpus passages; versioned, human- and machine-readable schema (JSON + Markdown) with an inspectable change history. (c) Evaluation set: curated test cases with documented target judgements for measuring the quality of every later use.
§2 Dialogue instance. Executable configuration (system instructions, retrieval architecture over §1, citation logic, access and response guardrails) that makes the knowledge base queryable in dialogue via an AI foundation model; answers reference the model components used. Provided either hosted (logically tenant-separated) or as a transferable configuration for your own infrastructure. Foundation-model-agnostic: §1 is separate from any specific AI model and portable to successor models.
§3 Interfaces. Documented REST API and MCP server configuration (Model Context Protocol) for integration into the client's systems; access keys held in the client's custody.
§4 Dossier. Editorially composed print version of the curated knowledge layer (“decision fingerprint”), PDF and bound copy.
§5 Rights and data. In §1, §2 (configuration) and §4 the client receives exclusive, perpetual, transferable and inheritable usage rights. Personal data is subject to the GDPR (access, export, erasure). Method, generic tools and question and evaluation templates remain the intellectual property of gannaca; the client receives the non-exclusive licence required to use their instance. No use of client data to train third-party models; no disclosure to third parties.
§6 Exclusions. Not owed: training or fine-tuning of a dedicated AI foundation model (can be commissioned separately); rights in third-party foundation models (their use is subject to the respective licences and can be substituted); the suitability of model-based statements as decision, legal, tax or investment advice — the dialogue instance is an assistance and memory system, decisions are made by the user. Availability according to a separate SLA.
What this looks like in everyday life.
“Would the senior have signed?”
11 pm, signing is tomorrow. The successor tests her decision against her father's model — and gets his test questions instead of a gut feeling. She spares herself the phone call.
The written second opinion
Every deal memo runs through the founder's model before the meeting. The committee sees: approval, concerns, and which experience episode the concern comes from.
Onboarding in three conversations
The new board member asks the model how the firm decided in 2009, 2020 and 2023 — three decades of context before the first meeting begins.
Handover without perpetual consulting
With the company, the buyer receives the former owner's judgement model. The knowledge transfer that otherwise takes three years of consulting contract is part of the asset.
In 2046, the granddaughter asks
How did grandmother think about risk when everyone got nervous? No second-hand anecdote — her decision logic, in her words, with her examples.
Consistency check against yourself
The decision-maker tests himself against his own model: where do I deviate from my own rules today — and is that growth or drift?
Your model works
where you need it.
A Mind Model that lives only in an app would be half the idea. That is why every model is designed from the outset as a connectable building block — for your family office, your company, your successors. Three levels, ascending by depth of integration:
A·01Dialogue — usable immediately, without an IT project
From handover, your model is queryable as a protected counterpart: in the familiar messenger, via private web access, with your voice if desired. Advisory board, family or successors ask questions — the model answers in your decision logic and discloses which of your experience statements it relies on.
A·02Integration — the model in your systems
- API access: your organisation queries the model programmatically — as a second opinion in investment committees, say, or as a review instance in deal pipelines.
- Open standards: connection to common enterprise AI environments via open protocols (including MCP) — the model appears there as a tool alongside your other systems.
- Agent embedding: virtual employees that pre-screen, prioritise and escalate with your judgement logic — the human decides, the model prepares.
A·03Sovereignty — ownership, operation, control
- Ownership: model, raw data and condensations belong to you — contractually, completely, exclusively. mindmodels retains nothing.
- Operation of your choice: hosted and tenant-separated — or handed over into your private infrastructure (on-premises/private cloud).
- Model independence: the knowledge base is kept separate from the respective AI foundation model; when the model generation changes, your Mind Model moves with it — it does not age with the technology.
- Versioning & erasability: every state is reproducible, every layer can be erased without residue on request (GDPR-compliant, access and export included).
Two ways in — one costs €490, the other a conversation.
The Probe is the small proof in seven days, fully credited towards Tier I. The mandate is the whole way: 30 days, one model, your property. For universities and organisations there is the Cohort Programme.