AI that has to survive reality.
Agentic systems, AI architecture, product thinking and the unglamorous work of turning demos into things organizations can actually use.
I like systems. I like learning how they work, pushing them, measuring them, then rebuilding them better.
I’m Amr ElSehemy — founder of Knowlly and a Senior Manager at Deloitte Innovation Hub, working across AI, data, engineering and enterprise innovation.
But titles only explain part of it. I’m happiest when I’m turning an idea into something real: an AI system, an app, a smart-home setup, a stubborn shutter, a piece of furniture, or a physical system I want to understand well enough to improve.
The common thread is curiosity with execution. I don’t enjoy observing systems from the outside. I want to get inside them.
Agentic systems, AI architecture, product thinking and the unglamorous work of turning demos into things organizations can actually use.
My company, my sandbox and my vehicle for building useful systems without waiting for permission.
HYROX, running, wearables, lactate, recovery, pacing, wall balls — a constant experiment in making the machine better.
LEGO, Bosch tools, DIY, smart-home systems and hands-on fixes all scratch the same itch: understand how the parts work, then make the whole thing better.
Surface knowledge bores me. Whether it’s agent architecture, HYROX pacing or a power tool, I want enough understanding to form my own opinion.
→Ideas get more interesting when they become software, systems, prototypes, dashboards, routines or physical things you can touch.
→Data is useful when it changes a decision. I measure performance, products and processes so I can see what actually works.
→I’m rarely finished. The fun is in the next iteration.
→Software, Microsoft technologies, mobile, cloud and years of building before “AI” became everyone’s headline.
Leading, building and advising around AI, data and agentic systems while growing Knowlly alongside it.
Products, inventions, experiments, races, homes, systems — whatever earns the obsession next.
This site is where I document the useful parts — what I’m building, learning, testing and occasionally getting wrong.
Read the notebook →