After 30+ years coding and 8 years in AI, here’s my bet on what comes after LLMs: agents that don’t just respond but actually *do* things autonomously.
The story
After 30+ years writing code and eight years deep in AI, I keep hearing one question: “What comes after LLMs?”
Here’s what I’m betting on: AI agents.
LLMs are brilliant at understanding and generating text. But they’re fundamentally passive – they respond, they don’t *do*. When you close the chat, they forget you existed.
Agents flip that script entirely.
Instead of waiting for instructions, they pursue goals. They can draft your email, send it, track the response, schedule a follow-up, and update your CRM. All without you hovering over them.
The shift is from intelligence to *agency*.
What makes this transition matter:
LLMs excel at language. Agents add autonomy, persistent memory, and the ability to use tools. They don’t just answer questions – they solve problems across multiple sessions and systems.
Skills that’ll separate the winners:
– Task decomposition – breaking goals into safe, executable steps agents can handle
– Orchestration – connecting agents to APIs, databases, and workflows that matter
– Oversight – monitoring autonomous systems that can surprise you
– Iterative optimization – refining agent behavior through feedback loops
Frameworks like AgentQ and Nvidia NIM show we’re past the proof-of-concept phase. The challenge now isn’t building smart models – it’s managing workflows, context switching, and persistent memory at scale.
Next up? Agent ecosystems. Coordinated swarms working across enterprises, adapting and delegating with minimal human input.
The people who master designing, deploying, and managing these systems won’t just use AI differently
Recent Comments