May 11 · 5 min read · Silent failures kill AI agents in production. They don’t crash. They don’t throw errors. They just stop doing what you trained them for. This is not a corner case — it’s the default failure mode. I’m calling this pattern Agentic Drift — the gradual, ...
Join discussion
May 8 · 5 min read · Multi-agent systems are stuck. The agents themselves—LLMs, microservices, tools—are no longer the bottleneck. The problem is orchestration: the missing contract layer that guarantees coordination, discovery, updates, and compliance at scale. Without ...
Join discussion
May 7 · 5 min read · --- title: "AI Mode in Chrome Is Not Assistantware" description: "Chrome’s AI Mode is augmentware, not assistantware—Google’s quiet retreat reveals a fundamental product architecture truth." date: 2026-04-17 categories: ['AI Product Design', 'Agentic...
Join discussion
Apr 24 · 5 min read · Everyone building AI systems right now has the same instinct: if one agent is good, multiple agents debating must be better. It feels rigorous. It feels like ensemble methods from classical ML. It is a trap. The consensus problem nobody talks about M...
Join discussion
Mar 19 · 8 min read · Every production agent system I've worked on has the same failure mode. Context rot. Stale artefacts silently served to the model. No audit trail for what was included or excluded. Token budgets blown with no graceful recovery. Multi-agent context bl...
Join discussion
Mar 19 · 5 min read · The OS-Paged Context Engine handles the technical lifecycle: what loads, what gets evicted, what passes validation. It produces an immutable manifest for every call. But the manifest tells you what the model saw. It does not tell you whether it shoul...
Join discussion
Mar 18 · 7 min read · Most companies using AI are doing substitution. Replace a copywriter with GPT-4o. Replace a data analyst with a BI copilot. Replace support agents with a chatbot. These are real productivity gains. They are not compounding. The distinction matters be...
Join discussion
Feb 28 · 4 min read · OpenCode issue #10704 landed this week: "Use provider-hosted web search when available." The request is specific and correct. Web search in most AI agents today is hardcoded: Exa integration, custom flag, API key. Even when the provider — Claude, Ope...
Join discussionDec 2, 2025 · 9 min read · What if your software testing framework could evolve from a static, rules-driven process into a dynamic, adaptive system empowered by agentic AI use cases and agentic AI applications that continuously optimize workflows? Instead of executing repetiti...
Join discussion
Nov 6, 2025 · 14 min read · Introduction The future of software testing isn't just automated—it's agentic. While traditional test automation requires developers to write explicit instructions for every scenario, agentic QA systems use AI to understand, reason, and adapt their t...
Join discussion
