Deep technical discussions on AI agents, frameworks, and software architecture.
I put together a development guide for the Vision Pro and the biggest lesson is that spatial computing is not '3D UI.' It's a different input model, a different ergonomics problem,
I wrote a piece on quantum's practical applications this year and tried to stay grounded about it. We're not replacing classical computing, but specific problems optimization, chem
The agent framework space right now is chaos in the best way. NVIDIA dropping OpenCLAW into the mix made me sit down and write up my take on where it's all heading. My honest read
I wrote up some thoughts on the Claude agentic revolution and honestly the biggest change I notice is the default: models now assume they'll use tools and take actions, not just an
Support engineering is one of those roles people assume AI will just delete. In practice, we're seeing it shift from 'tier-1 triage' to 'people who supervise and improve the agents
Arabic voice is one of those areas where 'good enough' isn't. Dialects, code-switching, and the sheer variety make automated customer service genuinely hard to get right. The blog
If you've ever tried to build anything with Arabic text, you know the struggle. Morphology, diacritics, dialect variation — it's a whole different beast from English NLP. We hit t
Wrote a longer piece on this, but the tl;dr is that moving inference onto the device changes everything — latency aside, it's the only clean answer to the privacy problem. For a l
'Vibe coding' is a fun term but I think it misses the point. You're not just chatting and hoping; you're directing, reviewing, and debugging at a higher level of abstraction. The
There's this narrative that prompt engineering is dying because models got smarter. I don't buy it. The gap between people who get mediocre outputs and people who get exceptional o
I've been meaning to share this for a while: how I structure AI-powered web apps on Next.js, especially when you want streaming, edge rendering, and a solid DX. The short version
I keep coming back to this: the real productivity win isn't a smarter chatbot, it's an agent that can execute a whole task end to end. We noticed it clearly while building Fahhim —
I wrote a piece a little while ago about how 2025 was the year of AI assistants and 2026 is turning out to be the year of agents. The more I build with them, the more I believe the
