Our AI Charter, an open approach for SMEs

At Bosqweb, we use artificial intelligence to make legacy systems easier to understand, reduce complexity, and help teams make better decisions. Instead of keeping our AI usage rules in an internal document, we chose to publish an open charter: you can reuse it, adapt it, and apply it to your own organization.

Open charter for responsible AI use in SMEs, v1.0

Preamble

Artificial intelligence can help organizations better understand their data, processes, and risks. It can also create confusion, dependency, or false certainty if used without a clear framework.

This charter proposes simple principles to guide AI usage in small and medium-sized businesses. It may be reused, adapted, and extended freely, provided the original source is acknowledged.

You are free to reuse, adapt, and cite this charter, as long as you reference “Open AI Charter for SMEs, v1.0 (inspired by Bosqweb)” or an equivalent mention.

1. General principles

1. AI should clarify, not obscure.
AI systems are used to clarify an organization’s situation (data, risks, options), not to hide complexity behind black boxes or jargon.

2. No unrealistic promises.
AI tools are not presented as magic. No guarantee of outcomes (profits, performance, timelines) is made without explaining assumptions, risks, and limitations.

3. Transparency about AI’s role.
Each meaningful AI use is identifiable: what it does, what data it relies on, and whether it advises, automates, or alerts.

4. Final decisions remain human.
Critical decisions (financial, HR, legal, security) are made by people, with full awareness. AI is a support tool, not the decision-maker.

5. Respect data and people.
Data is collected and used proportionately, securely, and in compliance with applicable laws. Impacts on employees, customers, and partners are considered.

6. Right to test and to stop.
AI projects can be limited in time or scope (pilot, proof of concept). If results are not good enough or introduce risk, the organization may stop or redesign the project.

2. Data and models

7. Data minimization.
We avoid collecting or storing more data than necessary for the AI tool to work. Retention periods are defined and reviewed.

8. Transparency about data flows.
Whenever possible, the organization knows where data goes, which providers process it, and whether it is used to train other models.

9. Awareness of model limitations.
Known limitations (error zones, potential bias, lack of data, industry-specific context) are explained in clear terms to decision-makers and users.

3. Organization and people

10. Team involvement.
People directly concerned (operations, IT, business teams) are informed about AI deployment and, when possible, involved in design or evaluation.

11. Upskilling, not abrupt substitution.
AI is used to augment teams (analysis, automation of repetitive tasks, better visibility), not to make them instantly obsolete.

12. Minimum user training.
Users receive basic guidance on what the tool can and cannot do, how to use it, and how to report an issue or inconsistency.

4. Impact, monitoring, and improvement

13. Measurable objectives.
An AI project targets concrete, observable outcomes (time saved, fewer errors, better visibility, improved service quality), even if measurements remain approximate.

14. Monitoring unintended effects.
The organization stays alert to side effects: workload shifting elsewhere, new errors, misunderstandings, perceived unfairness. These signals are taken seriously and may trigger adjustments.

15. Charter revisions.
This charter can and should be reviewed periodically based on experience, user feedback, and regulatory and technical developments.

5. Using and adapting the charter

Any organization, team, or person may:

• reuse this charter as-is,
• modify it to fit their context,
• cite excerpts in their own documents.

We only ask that you mention the source (“Open AI Charter for SMEs, v1.0, inspired by Bosqweb”) and indicate changes when the charter is substantially rewritten.