https://jamal.dev/tags/ai/Recent content in Ai on Jamal YusufHugoen-usMon, 18 May 2026 00:00:00 +0000https://jamal.dev/writing/production-ai-agents-enterprise/Mon, 18 May 2026 00:00:00 +0000https://jamal.dev/writing/production-ai-agents-enterprise/<p>An agent that cannot reach production systems is not an agent. It is a <strong>demo with ambition</strong> — and in healthcare, ambition without integration is just risk in a nicer font.</p> <p>I learned this leading multi-agent platforms wired into claims, membership, and payment flows — real latency SLAs, real audit requirements, real members on the other end of every invocation.</p> <h2 id="integration-over-isolation">Integration over isolation</h2> <p>The value was never &ldquo;another tab with a chat box.&rdquo; Experts wanted help <strong>inside the flow of work</strong> — grounded in operational context, historical data they could defend, and APIs that did not lie.</p>https://jamal.dev/writing/why-enterprise-ai-agents-fail/Thu, 15 Jan 2026 00:00:00 +0000https://jamal.dev/writing/why-enterprise-ai-agents-fail/<p>I have watched smart teams ship impressive agent demos — and then watch those same agents fail the only test that matters: <strong>would an expert trust this under pressure?</strong></p> <p>The model is rarely the villain. The retrieval design is.</p> <h2 id="the-expert-retrieval-gap">The expert retrieval gap</h2> <p>Most enterprise AI agents are built around a comforting pipeline: chunk, embed, retrieve, generate. It looks scientific. It scales on slides. It also assumes experts think in paragraphs — flat, interchangeable, equally worthy of attention.</p>