mistrah.com · Medium · LinkedIn · X · ORCID
I help leadership teams turn AI ambition into measurable, trusted results.
A strategy for AI is not a list of use cases. It is a decision system: where to focus, who decides, what to measure, what to refuse, and how a pilot becomes a run.
Entrepreneur based in Paris. The repositories below are the open-source side of my work: tools that make AI portfolios governable and their value provable.
| Project | What it does |
|---|---|
| strategy-os | Transformation Intelligence & ROI: design strategy, track execution, measure impact, attribute value and prove ROI. |
| scale-gate | Enterprise AI evidence and scale decision engine: from pilot promise to defensible scale. |
| radarline | Local-first Technology & AI Intelligence OS: from information noise to decision-ready signals. |
- It’s Not the Technology · SSRN 7038158 · doi:10.2139/ssrn.7038158
- Eighteen Frameworks, No Evidence · SSRN 7232782 · doi:10.2139/ssrn.7232782
- OpenAI's Cyber Model Finds Exploits at Scale. The Number That Matters Is Whether It Stays in Scope. · 9 Nov 2026
- The Best AI Model Scores 12% on Real Expert Work. Build Your AI Roadmap on That Number. · Towards AI · 3 Nov 2026
- The Era of Unlimited AI Money Is Over. Cheap AI Will Be Paid For, One Way or Another. · 2 Nov 2026
- Amazon and JPMorgan Are Blocking AI Agents. Walmart and Google Are Writing the Rules. · Towards AI · 2 Nov 2026
- The Pentagon Banned Claude. A Claude Agent Reportedly Ran on Its Network for Two Months Anyway. · 31 Oct 2026
- Mistral Large 4 n’est que 36e au classement : la souveraineté ne se gagne pas au score · 31 Oct 2026 (FR)
- Start here (FR): De l’ambition IA aux résultats mesurables : 5 décisions de gouvernance
L’IA n’est pas un outil. C’est une décision de leadership. · Abbas Mistrah · Paris