Product

Heisterlux is being built as software for understanding organizational reliance on AI: what is used, what may depend on it, and what a change could mean.

Heisterlux is in pre-launch. Nothing below is generally available yet. The roadmap shows the real status of every capability.

What it is being built to do

Visibility

Which AI systems and AI-enabled services are in use, where, and through which access route.

  • AI inventory

    Status: In development

    Record which AI systems and AI-enabled services your organization uses, per location.

    Scope
    Organization-level register of AI systems, capabilities and the services they appear in.
    Built so far (internal)
    An asset register exists in the product codebase and runs in a development environment for internal testing.
    Still to build
    Customer onboarding, a production release, a support route and approval to publish.
    Customer use today
    Not available
    Last reviewed

    Limitations:

    • In the current scope, entries are recorded by people in your organization. Automatic discovery is not part of it and is not promised.
  • Deployment Contexts

    Status: In development

    Record how your organization obtains access to an AI capability — for example an organization-managed licence, an individual subscription or an API agreement.

    Scope
    Describes the access route separately from the AI capability itself and from the application it appears in.
    Built so far (internal)
    A Deployment Context model exists in a development environment. It has not been released to production.
    Still to build
    A production-ready release path, customer onboarding, a support route and approval to publish.
    Customer use today
    Not available
    Last reviewed

    Limitations:

    • The host application (such as a word processor or chat tool) is recorded separately; it is not a Deployment Context.

Reliance

What the organization depends on them for, how critical that is, and why it was considered acceptable.

  • Reliance and criticality

    Status: In development

    Describe what your organization relies on an AI-enabled service for, how critical that is, and why the reliance was considered justified.

    Scope
    Per use: the decision or task supported, its criticality, and the recorded rationale.
    Built so far (internal)
    An internal foundation for recording organizational uses, their original justification and criticality declarations exists in a development environment. Parts of the recording screens exist there too.
    Still to build
    Customer-ready recording flows, verification of who may declare what, a production release, a support route and approval to publish.
    Customer use today
    Not available
    Last reviewed

    Limitations:

    • Criticality is an organizational judgement that Heisterlux records; it does not assign it for you.

Change

What shifted: a model version, the terms of access, the application around it, or the way it is used.

  • Review signals

    Status: In development

    Flag recorded reliances that may need a fresh look because something about them changed or aged.

    Scope
    Signals derived from information recorded in Heisterlux.
    Built so far (internal)
    A review-signal view exists in a development environment.
    Still to build
    Evaluation against real customer data, a production release, a support route and approval to publish.
    Customer use today
    Not available
    Last reviewed

    Limitations:

    • Signals depend on what has been recorded. Heisterlux does not yet watch vendors for changes automatically.
    • A signal is a prompt to look, not a finding that something is wrong.
  • External change sources

    Status: Exploring

    Bring in published vendor changes — such as model retirements or terms updates — as input for review signals.

    Scope
    Curated, source-linked change records.
    Built so far (internal)
    Nothing built so far.
    Still to build
    A decision on whether and how published vendor changes can be brought in reliably. Sources, coverage and accuracy are open questions.
    Customer use today
    Not available
    Last reviewed

    Limitations:

    • Exploring. We may not build this, or not in this form.

Impact

Which applications, decisions and reliances a change may affect — and where the evidence is not enough to say.

  • Impact view

    Status: In development

    When something changes, show which recorded uses may be affected — and where the evidence is not enough to say.

    Scope
    From a recorded change to the justification items and uses that depend on it, with uncertainty shown.
    Built so far (internal)
    An internal foundation determines, for directly recorded justification items, whether their support changed, records a direct affected determination and derives a remaining-support view. It is tested internally and exists in a development environment only.
    Still to build
    Indirect or inferred impact is not built. There is no customer-facing surface yet. Evidence limits and human review have to be designed in before a production release.
    Customer use today
    Not available
    Last reviewed

    Limitations:

    • A dependency path or a model change does not by itself show that a decision is affected or no longer valid. People make that judgement.
    • Only directly recorded items are evaluated today.

Action

A suggested next step that fits the size of the change and the stakes of the reliance. People decide what to do.

  • Proportionate next steps

    Status: In development

    Suggest a next step that fits the size of a change and the criticality of the reliance — for example reconfirm, reassess or escalate — based on recorded facts and declarations.

    Scope
    Rule-based recommendations derived from recorded facts, declared criticality and the state of the supporting evidence. A recommendation is a suggestion: not a decision, and not enforced.
    Built so far (internal)
    A deterministic, rule-based recommendation foundation (no language model) has been tested, promoted in the code repository and activated in a development environment. It is internal only.
    Still to build
    A customer-facing explanation and surface, a way for customers to declare criticality and record decisions with verified authority, a production release, a support route and approval to publish.
    Customer use today
    Not available
    Last reviewed

    Limitations:

    • Missing information produces "insufficient basis", not a conclusion.
    • Heisterlux does not provide legal advice, and recommendations are not legal or regulatory requirements.

Underneath

What makes the other steps explainable later.

  • Evidence trail

    Status: In development

    Keep a record of who recorded or changed what, and when, so reliance decisions can be explained later.

    Scope
    Change history for recorded information.
    Built so far (internal)
    Append-only recording with history exists in a development environment. It has not yet been independently verified as a customer-facing audit trail.
    Still to build
    Independent re-test, customer-facing history views, then a production release, a support route and approval to publish.
    Customer use today
    Not available
    Last reviewed

    Limitations:

    • Not a certified audit log.

See every capability and its status

Who it is for

  • People responsible for how AI is used across an organization — in IT, risk, compliance, information security, data protection or operations.
  • Organizations where AI appears in many tools and teams, not only in one central project.

What it is not

  • Not a monitoring agent. The current scope relies on information people in your organization record; it does not scan devices, networks or employee activity. Automatic discovery is not part of the current scope and is not promised.
  • Not a compliance certificate: it helps you understand and document reliance; it does not certify it.
  • Not legal advice: decisions stay with your organization and its advisers.