Get started for free
Company

Founder Memory Governance Standard

What Zero→1 remembers about your founder journey, why, who can see it, and the boundaries that keep it yours.

Version
1.0
Status
Company-wide standard
Applies to
All founders

1. Why This Standard Exists

Founder Memory exists to maintain useful continuity across the Founder Operating System. It allows Founder Intelligence to understand relevant founder context over time without requiring you to repeatedly reconstruct your business, evidence, decisions and progress.

The governing architectural decision is:

Founder Memory is a curated operating state, not an indiscriminate record of everything the founder has ever said or done.

Information should earn persistence because retaining it has a legitimate and continuing purpose in helping you operate.

More memory is not inherently better memory.

2. The Principles It Lives Inside

Private by default. Founder Memory forms part of your private operating environment. It is not automatically visible to other founders, programme organisations, sponsors, institutions, employers or other external parties.

Founder Intelligence recommends. Founders decide. Founder Memory informs Founder Intelligence. It must not give AI-generated assumptions greater authority than your decisions or actual evidence.

Evaluate the work, not the founder's worth. Founder Memory should primarily describe the business, evidence, decisions, assumptions, actions and progress rather than unnecessarily constructing judgements about you as a person.

Capability over dependency. Founder Memory exists to increase useful continuity and founder capability, not to create psychological dependence on Zero→1.

Evidence is the operating currency. Founder Memory must preserve meaningful distinctions between what you believe, what evidence demonstrates, what you have decided and what AI has inferred.

3. What Founder Memory Is, and Is Not

Founder Memory is the governed, persistent and version-aware operating record of relevant founder context, evidence, decisions, commitments and other information required to maintain useful continuity across the Founder Operating System.

Founder Memory is not:

These distinctions are reflected in the underlying architecture, not merely in legal documentation.

4. Founder Content and Founder Memory Are Different

Founder Content is the wider body of information you create, enter, upload or supply: Mission responses, Founder Intelligence conversations, documents, images, audio, video, research, evidence, plans, notes, reflections and business information.

Founder Memory is the curated operating state derived from or linked to relevant information.

For example, you may upload a 50-page customer research report. The report remains Founder Content. Founder Memory might appropriately contain:

Where appropriate, those memory items retain provenance back to the underlying source.

Access to Founder Content does not mean all Founder Content becomes Founder Memory.

5. The Memory-Worthiness Test

Information does not become persistent Founder Memory merely because the system has encountered it.

Before creating or materially retaining a memory, the system effectively asks:

Does remembering this materially improve the founder's ability to operate, make progress or receive useful continuity from the Founder Operating System?

Appropriate candidates include your current business direction, customer hypothesis, problem hypothesis, value proposition, business model, pricing hypothesis, evidence state, significant test results, important decisions, commitments, Mission outcomes, unresolved material assumptions, current operating priorities and material commercial constraints.

Information with no continuing operating relevance, such as incidental remarks, temporary moods, unrelated personal interests or conversational detail, does not automatically become persistent memory.

Persistence must be purposeful, not opportunistic.

6. Every Memory Knows What Kind of Information It Is

Founder Memory preserves meaningful distinctions between:

These do not silently collapse into one category of truth.

"Customers will pay £49 per month" may be your hypothesis. It must not automatically become an evidenced fact.

"Founder appears reluctant to conduct customer interviews" is an AI inference. It must not silently become "founder avoids customer research" as a permanent characteristic.

The authority of a memory must reflect what kind of information it actually is.

7. Provenance

Material Founder Memory carries appropriate provenance: where the information came from, when it originated, whether it came directly from you, from evidence, from the system or from AI inference, and where the underlying source can be found where applicable.

No material Founder Memory without appropriate provenance.

Not every piece of provenance is displayed during ordinary use, but it exists in the architecture to support accuracy, explainability and governance.

8. Versioning, Not Silent Overwriting

Important Founder Memory evolves rather than being silently overwritten.

The current state is Version 2. Version 1 remains historically relevant without continuing to operate as your current direction.

This matters most for direction, customer, problem, business model, pricing, important assumptions, commitments and material decisions.

Current state and historical truth are different things. Where useful, Founder Intelligence understands both.

Memory can be current, superseded, historical, disputed, uncertain, temporary, or deleted and no longer available to Founder Intelligence.

9. Founder Decisions Have Authority

Founder Intelligence may challenge a decision. It must not silently replace it.

Suppose Founder Memory contains "founder decision: target independent consultants" and Founder Intelligence later concludes the evidence is stronger among SMEs. The appropriate behaviour is to challenge you:

"Your current direction is independent consultants, but the evidence you've gathered appears stronger among SMEs. Do you want to reconsider the customer direction?"

If you change direction, a new founder decision supersedes the previous one.

AI inference must not silently override a founder decision.

10. Evidence Can Challenge Existing Memory

Founder authority does not mean Founder Memory ignores contrary evidence.

Suppose your hypothesis was "customers will pay £49 a month", and 0 of 15 customers accepted it. Founder Intelligence may appropriately say:

"Your original pricing hypothesis was £49 a month, but the evidence gathered since then does not currently support it."

The historical hypothesis remains accurate as a record of what you believed. Its evidential status has changed. The architecture keeps "what the founder believes or decided" distinct from "what the evidence currently supports".

11. AI Inference Has Limited Authority

AI-generated inference remains identifiable as inference. Founder Intelligence does not become an invisible authority capable of manufacturing persistent "facts" about founders.

Where material, an inference carries its basis, source, date, current status and uncertainty, and whether later evidence supersedes it. We avoid false numerical precision; states such as tentative, supported or strongly supported are preferred.

12. You Can Correct It

You can correct materially inaccurate operating context. Saying "that's wrong", "that has changed", "you've misunderstood me" or "I changed direction" affects Founder Memory appropriately.

Correction does not necessarily require historical deletion. "Previous direction: B2C. Superseded by founder decision: B2B. Current direction: B2B" is often more accurate and useful than erasing the original state.

Where you materially dispute an AI inference, it must not continue operating as though undisputed. If the system inferred you were avoiding customer research and you reply "that's wrong, I've completed 20 interviews outside Zero→1", the inference is marked disputed, updated or superseded, and reliance on it is reduced or suspended.

Founder Memory should evolve with the founder rather than trapping them inside an outdated profile.

13. Sensitive Information Is Not Memory by Default

You may disclose information concerning physical health, mental health, ethnicity, religion, political beliefs, sexuality, disability, family difficulties, severe financial distress or other sensitive aspects of your private life.

The mere fact that information appears in a Founder Intelligence conversation does not justify converting it into persistent reusable Founder Memory. Intentionally persisting sensitive information requires a defined founder-benefit purpose, necessity, proportionality, lawful processing, appropriate security and approval where required.

Sensitive information disclosed incidentally must not automatically become persistent Founder Memory.

Commercial business information is different. Revenue, costs, runway, pricing, margins, customer concentration, funding requirements, forecasts and financial constraints may legitimately form part of Founder Memory: they are central to operating a business, and they are protected accordingly.

14. Purpose Separation and the Advertising Boundary

Founder Memory exists for the Founder Operating System. It does not automatically become a universal profile available for unrelated uses: advertising, remarketing, audience creation, personalised sales pressure, unrelated marketing segmentation, programme surveillance, institutional assessment or employer assessment.

Founder Memory, Founder Content and Founder Intelligence conversations must not automatically feed Meta advertising profiles, Google advertising profiles, lookalike audiences, behavioural remarketing or equivalent systems. This is a hard boundary, and the future introduction of advertising technology does not alter it.

The existence of Founder Memory for Founder Intelligence does not create permission to use that memory for another purpose.

15. Programme Organisations Do Not See Your Memory

Founder Memory is not programme reporting data by default. An organisation purchasing, funding or sponsoring your access does not thereby acquire access to it.

Programme organisations do not automatically receive information such as "Founder Intelligence believes founder is risk-averse", "founder has limited financial runway" or "founder is considering abandoning the business".

Appropriately defined programme progress or outcome information may be available under our separately governed Programme Visibility Standard.

Founder Memory is private operating context, not an institutional surveillance layer.

16. Company and Provider Access

Private by default continues to apply inside the company. Founder Progress Ltd processes Founder Memory where legitimately necessary to provide Zero→1, operate Founder Intelligence, provide support, maintain security, investigate technical failures or comply with law. Technical ability to access Founder Memory does not equal organisational permission to browse it.

Authorised service providers may process Founder Memory to deliver Founder Intelligence. Provider use is controlled: we track which providers receive what, why, their retention, their model-training treatment and their deletion arrangements.

Our engineering preference extends data minimisation into AI orchestration: send the context required for the task, not the founder's entire historical memory merely because it exists.

17. Model Training

Founder Memory exists to provide continuity to you. Its existence does not automatically authorise its use for external foundation-model training, unrelated model training, provider product improvement, advertising models or unrelated profiling models.

Any proposed training use involving Founder Memory or substantive Founder Content requires separate product, privacy and legal assessment before implementation.

18. Retention and Account Deletion

Memory is not architected as permanent. Different information carries different lifecycle treatment: current operating memory, historical versions, temporary inference, source content, legally retained records and telemetry are distinguished.

When you validly delete your account, we do not retain an active AI-readable Founder Memory profile merely because separate records must be kept for legal purposes. A transaction record required for accounting does not justify retaining your AI-derived operating profile and business thinking.

Deletion architecture distinguishes legal record retention from continued Founder Intelligence memory.

19. Transparency and Your View of Memory

You should be able to understand what Founder Memory does:

"Zero→1 remembers relevant information from your founder journey, such as decisions, evidence and your current business direction, so Founder Intelligence can give you useful context over time. It doesn't treat everything you say as permanent memory, and your private Founder Memory isn't automatically shared with programme organisations or other founders."

We preserve the ability to give founders a meaningful view of what Zero→1 currently understands about their business, organised around direction, customer, problem, evidence, business model, pricing, key decisions, assumptions, commitments and progress, with interactions such as "update", "that's changed" and "this isn't right".

20. Younger Founders

This Standard originated from our children's privacy work but applies to all founders. We do not build a separate, substantially inferior memory architecture for 16 and 17-year-olds; the company-wide architecture provides strong protection for everyone.

Governed contextual Founder Memory is intrinsic to the Founder Operating System. That never justifies unrelated profiling:

21. Questions

Read this Standard alongside our Founder Intelligence Agency Standard, Programme Visibility Standard, AI & Founder Guidance Transparency Statement and Privacy Notice.

Questions or concerns: support@zeroto1.co.uk

Founder Progress Ltd
trading as Zero→1
Company number: 17333996

UpdatesOctober 2026