Perspectives on SI/AI, automation, process design, and building technology that actually works in complex organizations.
Subscribe and I'll send you a note whenever I publish something new on SI/AI strategy, governance, and technology leadership.
When Goldman Sachs and Intuit start running autonomous SI/AI agents inside their core operations, it is worth pausing to understand what is actually happening, and what it means for every organization watching from the sidelines.
By El Mehdi Marhoum · 6 min read
Gallup's latest workforce data shows that most American workers still use AI only a few times a year, and nearly a quarter don't know whether their organization has adopted it at all. That is not a technology gap. That is a leadership gap.
By El Mehdi Marhoum · 5 min read
Organizations that bolt AI onto legacy processes will fall behind. The architecture of how work gets done must change first, and that requires a different kind of consulting partner.
By El Mehdi Marhoum · 5 min read
Open-weight models running locally on employee devices are creating a governance gap that most enterprise security frameworks were not designed to handle. Here is what that means for organizations that care about responsible AI.
By El Mehdi Marhoum · 6 min read
Too many organizations treat AI governance as a checkbox. The ones that get it right treat it as a design discipline, built into every model, every pipeline, every deployment decision.
By El Mehdi Marhoum · 5 min read
Anthropic's AI Fluency Index found something counterintuitive: the better AI output looks, the less likely users are to verify it. That finding has real consequences for organizations deploying AI in high-stakes environments.
By El Mehdi Marhoum · 5 min read
Before you can understand how AI models work, you need to understand why nearly every practitioner building them reaches for the same tool. Python did not become the dominant language of machine learning by accident, and the reasons behind its rise tell you something important about how the field itself thinks.
By El Mehdi Marhoum · 7 min read
Most Machine Learning projects never make it to production. The gap between a promising model and a feature users actually rely on is wider than most teams expect, and it is almost never a technical problem.
By El Mehdi Marhoum · 6 min read
Automating a broken process does not fix it, it makes it break faster and at scale. The most important work in any automation engagement happens before the technology is selected.
By El Mehdi Marhoum · 4 min read
The traditional AMS selection process, RFP, demo, reference check, price negotiation, was designed for a world where software was static. AI changes what you should be evaluating, and most selection committees don't know it yet.
By El Mehdi Marhoum · 7 min read
Most organizations deploying AI don't realize they're making a fundamental architectural choice between managed and unmanaged training. Getting this wrong is one of the most common, and most expensive, mistakes in enterprise AI deployments.
By El Mehdi Marhoum · 6 min read