Creator of DUBSAR · Author of Digital Trust
Sofiane Kotni
I design DUBSAR, a governance system for projects and automations assisted by AI agents.
The work addresses a practical question: how can teams preserve decisions, connect claims to evidence, expose contradictions and keep explicit human authority over sensitive actions?
DUBSAR
Govern AI-assisted work.
DUBSAR structures missions, constraints, evidence, limitations and human validations. It never gives an agent authority to approve its own work and does not promise automatic compliance.
The commercial path starts with a bounded audit: authorized sources, explicit coverage, evidence-linked findings and a report that the designated owner can validate.
This structure can support documentation readiness around the EU AI Act, without constituting legal advice, certification or an automated compliance verdict.
Background
From operations to evidence.
My background spans more than thirteen years across operational management, store leadership and regulated property diagnostics.
It taught me to define scope, observe facts, coordinate stakeholders, write reports and own decisions. That experience is the operational foundation of DUBSAR; it is not presented as thirteen years of AI experience.
Publication
Digital Trust
Digital Trust explores trust, judgment and responsibility in a world accelerated by artificial intelligence. The book is the editorial side of a line of inquiry put into practice through DUBSAR.
Profiles and support