Riley BrownJune 18, 202633m

9 AI Agent Trends To Get Ahead of 99% of People

TL;DR

Prompt hacks are fading fast: Brown says the old tricks like "I want you to act as" or manually @-mentioning files are becoming irrelevant because newer models in Claude Code, Codex, and GPT Image 2 respond better to plain English descriptions.
Skills are becoming the real primitive: He has created more than 100 skills in Codex, and instead of hand-writing them, he shows a faster loop where you ask the agent to do a task, refine the output, then turn that workflow into a reusable skill that can later auto-update itself.
AI agent tools are turning into super apps: Brown frames Codex, Claude chat, co-work, and Claude Code as general-purpose platforms that combine chat, coding, plugins, automations, hosted sites, multitasking, and an in-app browser in one place.
Agents are moving into Slack, iMessage, WhatsApp, and Telegram: Using Chorus, he demos cloud agents that never turn off, can reply in iMessage and Slack, analyze YouTube channels like Alex Hormozi and Theo from T3 Chat, build a public landing page, and email it to teammates.
The durable edge is not a tool, it is management skill: Brown argues that communication, delegation, taste, mental clarity, and multitasking matter more than any single platform, because managing agents will increasingly feel like managing a 24/7 team.
Model economics are splitting the market: Frontier systems like Fable are getting expensive, while open-source Chinese models are catching up. He highlights GLM 5.2 as "the best open-source model in the world" and recommends OpenRouter so users can route work to cheaper models when possible.

The Breakdown

Nine prompts on Fable 5 would have cost about $250 through the API, and Riley Brown argues that this cost curve, plus better open-source models and agents that can already control your computer, is why people need to become "agent native" now. His core point is simple: prompt hacks are dying, natural-language delegation is rising, and the winners will be the people who can clearly describe work, build repeatable skills, and manage fleets of agents across tools they already use.