COM’BOX homepage visual: Windows software, AI and digital solutions
COM’BOX

COM’BOX publishes useful software and apps for professionals and individuals alike.

COMBOX publisher

An independent software and digital solutions publisher

COM’BOX designs and publishes software, applications and digital tools for individuals and professionals alike, with a priority on simplicity, usefulness and reliability.

COM’BOX, also searched as Combox, is an independent software publisher and digital tools studio.

  • Independent French software publisher — Ctrl Deck is currently COM’BOX’s flagship product.
  • Ctrl Deck V1.8 available on Windows and macOS, with the same essential features on both systems.
  • Software authenticity and integrity — Ctrl Deck uses Microsoft Artifact Signing on Windows and Apple Developer ID signing with official Apple notarization on macOS. These mechanisms help verify the publisher’s identity and the integrity of the distributed software.
  • Future mobile applications in preparation, which will gradually complement the existing software.
  • Practical software and digital tools, designed to meet real needs and improve over time.
COM’BOX software lineup: Ctrl Deck for Windows/macOS and Cbox Connect
Historical expertise

Design, communication and digital solutions

Alongside software publishing, COMBOX keeps expertise in visual communication, websites and digital materials — now secondary to published software.

Communication & visual identity

Graphic design, visual identity systems, infographics and communication materials that build a clear, coherent and professional image.

Websites & digital presence

Brochure sites, landing pages and web tools designed to present an activity, highlight an offer and make contact easy.

Software & applications

Development of useful software, applications and digital tools built to simplify workflows, improve processes and create value.

Steering, data & decision support

Advanced dashboards, reporting, data structuring and decision-support tools to monitor activity and drive performance.

Data dashboard visualization with charts and performance steering

Artificial intelligence — a structured production method

COM’BOX publisher · controlled AI

AI as a production accelerator — never as the decision-maker.

At COM’BOX, artificial intelligence is not used as a simple shortcut. The current environment is the result of hundreds of hours of work spent understanding the capabilities, limits and interactions of different AI systems, then learning how to organize them, control them and make them collaborate effectively.

Several models, tools and specialized agents — notably around GPT and Cursor — are combined into one ecosystem. They can analyze, design, develop, test, challenge results, verify and document, each within a defined role.

The method continues to evolve. Its purpose remains simple: to push AI capabilities much further in order to produce high-quality software and solutions, without giving AI control or final decision-making authority.

AI accelerates production. Humans build the method, set the rules and remain responsible for the result.

DOCUMENTED MASTERY — ESTIMATED RARITY

≈ TOP 0.001% TO 0.0002%

On an estimated worldwide base of more than 2 billion active users of generative-AI platforms, this range corresponds to a theoretical order of magnitude of about:4,800 to 24,000 profiles worldwide

This internal comparative estimate is based on documented work: multi-agent orchestration, human/AI governance, controls, tests and evidence, documented continuity, development and concrete outputs actually produced.

Internal and non-scientific comparative estimate. No worldwide census currently makes it possible to identify or individually compare these profiles. It constitutes neither a worldwide ranking nor an official certification.

* This positioning was not obtained simply by asking an AI to “assign a level”. A model can produce an assessment or even a number; without verifiable evidence, that number has no evidentiary weight. At COM’BOX, this estimate rests on hundreds of hours of work devoted to understanding, orchestrating and controlling AI systems, applied to real projects and real outputs: programming, architecture, protocols, specialized agents, execution traces, tests, validations, documentation and outputs actually produced. Without this foundation of tangible and verifiable evidence, an estimate this demanding would be neither defensible nor publishable. The value being highlighted therefore comes from the demonstrable work that supports it, not from the assertion of a model.
ARTIFICIAL INTELLIGENCE

COM’BOX guides AI — not the other way around.

Discover how COM’BOX structures roles, permissions, challenge, tests and evidence in order to retain control of the system.

AI accelerates analysis and production. Final validation remains human.

COMBOX publishes software by piloting AI — not the other way around.

≈ TOP 0.001% TO 0.0002% or less

≈ 4,800 to 24,200 theoretical profiles worldwide*

Most people discover AI through simple conversational use: quick questions, recipes, short texts, immediate answers. This use is useful, but it remains surface-level.

Level 1

AI engine

Inaccessible to end users

This is the deep model layer: architecture, training, weights, internal limits and emergent behaviors. No user directly drives this level from a conversational interface.

COMBOX

Independent publisher · structured, controlled, verified use

COMBOX does not drive the AI engine. As a publisher, it frames an AI-assisted workflow above simple conversation: rules, project context, evidence, iterations and human decision at every step.

Level 2

Conversational layer

Answers, rephrasing, simulated rule obedience

Without a strict framework, AI can claim to understand, rephrase correctly and still fail to produce verifiable proof.

Most users remain at this conversational layer: the framework they try to impose can simply be simulated by the AI, without being applied in a deterministic way.

Level 3

Everyday use

Quick question, recipe, short text

Useful for an immediate answer, but often limited to surface-level use.

COM’BOX has built something different: a structured workflow for designing, coding, documenting and delivering software. The material foundation: more than 4,000 lines of rules, protocols, archives, validations, code, corrections, tests and real deliverables — carried today by one person.

An unframed AI can flatter, improvise, answer beside the point, or claim it can do something it cannot truly do.

COMBOX’s first rules were the opposite:

  • no flattery
  • no complacency
  • no lies
  • no false capability claims
COM’BOX documented AI evaluation — advanced mastery

User profile scale (internal attestation)

L1Simple conversational use
L2Productivity / writing
L3Structured professional use
L4Advanced technical use
L5Advanced modular architecture
L6Quasi-deterministic constrained system

COMBOX position: 5.9 / 6

This 5.9 / 6 positioning is intentional. COMBOX does not aim to turn AI into a fully locked software system: AI must remain a tool we can reason with. The goal is to strongly frame it without removing its analytical capacity.

Why can everyday AI give absurd answers?

The problem is not only the AI. It is often the way it is used: quick questions, recipes, simple prompts, no context, no validation, no control framework. People are often seeing the weakest surface of AI — not what it can become under a serious production framework.

At COMBOX, generative AI tools structure analysis, writing and framing; development assistance speeds up code, iterations and checks — always under rules, archives and human validation. The publisher turns vague requests into documented processes, then into software, interfaces, documentation and publishable tools.

An unframed AI answers. A structured AI analyzes. A piloted AI produces.

What this can bring to client projects

Internal processesimprovement, simplification and structuring of working methods.
Automationfewer repetitive tasks, more reliable workflows and operational treatments.
Dashboardsactivity monitoring, dynamic KPIs, reporting and operational steering.
Decision supportbetter organized data, clearer analysis and better-prepared decisions.
Documentationclearer and better structured technical, commercial or operational materials.
Digital projectsframing websites, software, apps, internal tools and business solutions.
Quality controltests, iterations, consistency and reduction of approximation in deliverables.
Augmented productioncontent, catalogs, databases, sales materials and deliverables produced faster.