Back to PREEXEC™
What How Why Compliance Domains Capabilities Use Cases Library Plans ROI Calculator Live Demo ↗
Use Cases · Eight Domains

Where structural clarity
becomes operational.

Wo strukturelle Klarheit
operativ wird.

Là où la clarté structurelle
devient opérationnelle.

Donde la claridad estructural
se vuelve operativa.

Dove la chiarezza strutturale
diventa operativa.

Yapısal netliğin
operasyonel hale geldiği yer.

Де структурна ясність
стає операційною.

حيث تصبح الوضوح الهيكلي
عملياً.

जहां संरचनात्मक स्पष्टता
संचालनात्मक बनती है।

PREEXEC™ is domain-neutral — the architecture is identical across all contexts. The policy defines what structural clarity means in each domain. These eight use cases show the architecture at work.

See it in a real workflow

A law firm lets its AI assistant work on client files — and every request is checked by PREEXEC before it runs.

MHV LEGAL — demo environment · AI-generated. EXECUTE / HOLD / BLOCK is decided before anything runs; every decision lands in a tamper-evident audit trail.
See the live demo →
Tax advisory — in a real workflow

A tax firm lets its AI assistant work on client files — and every request is checked by PREEXEC before it runs.

BMS STEUER — demo environment · AI-generated. EXECUTE / HOLD / BLOCK is decided before anything runs; every decision lands in a tamper-evident audit trail.
See the live demo →
Private banking — in a real workflow

A private bank lets its AI assistant work on client files — and every request is checked by PREEXEC before it runs.

MWP WEALTH — demo environment · AI-generated. EXECUTE / HOLD / BLOCK is decided before anything runs; every decision lands in a tamper-evident audit trail.
See the live demo →

Built for regulated, confidentiality-driven work

Tax Advisory & Audit

An associate drops client figures into an AI assistant to draft a memo.

Problem
§203 StGB makes unauthorised disclosure of a client secret a criminal offence — and §62a StBerG makes the firm responsible for the services it uses. Forbidding AI costs productivity; ungoverned AI is a liability.
PREEXEC™ evaluates
On-premise, before the AI acts: the clarity and policy-conformity of the request — and writes a tamper-evident audit entry. Client data never leaves the firm.
Outcome
Clear, in-policy requests run. Ambiguous or non-conforming ones are held or blocked — provably, for the Kammer or an auditor.
Private Banking & Wealth

An advisor uses AI to interpret a client order or draft regulated communication.

Problem
MiFID II suitability, MaRisk and DORA require demonstrable control over what AI is allowed to act on.
PREEXEC™ evaluates
A deterministic gate in front of every AI action — finance-calibrated, fully on-premise and auditable.
Outcome
Provable control over AI inputs: same input, same verdict, every decision recorded.
Law Firms & Litigation

An associate pastes case facts into an AI tool to draft a brief.

Problem
Legal professional privilege, §43e BRAO and §203 StGB turn an uncontrolled AI leak into a serious breach.
PREEXEC™ evaluates
Checks every request on-premise before anything runs; the verdict and its reason land in a tamper-evident audit trail.
Outcome
See it in the demo above — EXECUTE, HOLD and BLOCK decided before the AI touches a file.
Legal & Mediation

A mediator submits a settlement proposal into an AI-assisted negotiation system.

Problem
Legal inputs often contain emotionally charged language, incomplete clauses, or ambiguous party references — all of which can cause AI systems to misinterpret intent and produce flawed procedural outputs.
PREEXEC™ evaluates
Syntactic completeness of the clause structure, semantic unambiguity of the obligation terms, affective stability of the language, and conformity with the ADR policy ruleset. An immutable Audit Chain entry is created before any AI processing occurs.
Outcome
Structurally complete proposals pass to the AI. Ambiguous or emotionally unstable inputs are held for human review — with the specific structural deficiency logged in the Audit Chain.
HOLD on ambiguity
Financial Services

A trader inputs a natural language order instruction into an AI-driven execution system.

Problem
In regulated trading environments, structurally ambiguous or incomplete instructions can trigger incorrect executions — with consequences that cannot be undone post-execution.
PREEXEC™ evaluates
Syntactic precision of the order instruction, semantic unambiguity of quantity and instrument references, affective neutrality, and conformity with the Finance policy ruleset. Every evaluation is recorded in the Audit Chain with a SHA-256 hash before execution.
Outcome
Clear, complete instructions execute. Structurally incomplete or anomalous inputs are blocked — supporting traceability requirements relevant to regulated trading environments.
BLOCK on incomplete instruction
Healthcare & Medical

A clinician inputs a diagnostic query into an AI-assisted decision support system.

Problem
High-risk medical AI systems require human oversight and structural traceability. Incomplete or emotionally distorted clinical inputs can lead to dangerous diagnostic suggestions with no pre-execution safeguard.
PREEXEC™ evaluates
Structural completeness of the clinical query, semantic unambiguity of symptom and patient references, affective stability of the language, and conformity with the Medical policy ruleset. Every decision enters the Audit Chain before the AI model is reached.
Outcome
Structurally sound queries pass. Ambiguous or affectively distorted inputs are held for clinician review — supporting human oversight objectives with a verifiable Audit Chain.
HOLD for clinician review
Defense & Security

An analyst submits a situation assessment to an AI-assisted command support system.

Problem
In high-stakes operational contexts, structurally incomplete or emotionally distorted inputs to AI systems can produce catastrophically incorrect assessments. There is no second chance post-execution.
PREEXEC™ evaluates
Syntactic clarity of the assessment, semantic unambiguity of tactical references, affective stability under operational pressure, and strict conformity with the Defense policy ruleset. Every evaluation is recorded in a cryptographically signed Audit Chain.
Outcome
Only structurally sound, policy-conformant assessments pass to the AI system. All others are blocked or held — with a tamper-evident Audit Chain record for operational documentation.
BLOCK on structural failure
Education

A student submits an essay prompt to an AI-powered writing feedback system.

Problem
Educational AI systems process inputs from vulnerable populations. Emotionally distressed, manipulative, or structurally incoherent prompts can extract inappropriate responses or bypass content safeguards — with reputational and safeguarding consequences for institutions.
PREEXEC™ evaluates
Structural coherence of the prompt, semantic unambiguity of the learning objective, affective stability of the student input, and conformity with the Education policy ruleset — all content-agnostically, before any AI model processes it.
Outcome
Clear, educationally coherent prompts proceed. Emotionally distorted or structurally anomalous inputs are held for educator review — protecting students and institutions simultaneously, with every decision logged in the Audit Chain.
HOLD on affective distortion
Robotics & Autonomous Systems

An operator issues a natural language task command to an autonomous robot in a logistics facility.

Problem
Ambiguous natural language commands to robotic systems can produce physically irreversible actions. A structurally incomplete instruction — such as a missing object reference or contradictory spatial constraint — cannot be corrected once the robot has acted.
PREEXEC™ evaluates
Syntactic completeness of the task command, semantic unambiguity of object and spatial references, affective neutrality of the instruction, and conformity with the Robotics policy ruleset — deterministically, in milliseconds, before the command reaches the motion planner.
Outcome
Structurally complete commands execute. Ambiguous commands are blocked — preventing physical errors before they occur, with every rejected command recorded in the Audit Chain for operational review.
BLOCK on ambiguous command
Critical Infrastructure

An engineer submits a maintenance instruction to an AI-managed grid control system.

Problem
In energy, water, and transport systems, a single structurally ambiguous instruction to an AI control system can cascade into service disruptions or safety incidents. High-risk AI in critical infrastructure requires demonstrable pre-execution controls.
PREEXEC™ evaluates
Structural precision of the instruction, semantic unambiguity of system and component references, affective neutrality, and conformity with the Infrastructure policy ruleset. The Audit Chain creates a tamper-evident record of every pre-execution decision for regulatory inspection.
Outcome
Structurally precise instructions pass. Ambiguous or non-conformant instructions are blocked — providing a verifiable, continuous Audit Chain that supports regulatory inspection.
BLOCK on non-conformant instruction
Enterprise AI

An employee queries an internal AI assistant with a sensitive HR or compliance question.

Problem
Enterprise AI deployments face dual risk: structurally ambiguous inputs produce unreliable outputs, while emotionally charged inputs in sensitive HR or legal contexts can lead to outputs that create liability. Neither can be audited after the fact with sufficient granularity.
PREEXEC™ evaluates
Structural clarity of the query, semantic unambiguity of the subject matter, affective stability of the input, and conformity with the Enterprise policy ruleset — before any AI model responds. Every evaluation generates an Audit Chain entry for compliance review.
Outcome
Clear, structurally sound queries proceed. Sensitive or structurally ambiguous inputs are held for human review — giving enterprise compliance teams a complete, queryable Audit Chain without requiring post-hoc log analysis.
HOLD for compliance review

Note on use

PREEXEC™ is a deterministic measurement tool for evaluating AI inputs and outputs. It does not make autonomous decisions about persons, matters, or legal consequences. Verdicts (EXECUTE / HOLD / BLOCK) are technical classifications based on configured thresholds; operational and legal responsibility for decisions made using these classifications rests entirely with the system operator. Compliance reports, audit trails, and reproducibility evidence are documentation aids and do not replace qualified compliance assessment.