I'm building AI agents that do brand strategy. Not chatbots. Not prompt templates. Multi-agent systems where each agent has a specific strategic role.
The architecture
One agent runs research. Another maps category structure. A third generates positioning hypotheses. A fourth pressure-tests them.
They work together. They argue. They converge. It looks like a strategy team that never sleeps.
What I've learned
- Agents are only as good as the framework. Without a clear strategic method, AI produces plausible-sounding nonsense
- Specialization matters. One generalist agent is worse than four specialists
- The human still decides. Agents generate options. Judgment is still the strategist's job
Where this goes
The future isn't "AI replaces strategists." It's strategists with agent teams that compress weeks of research into hours.
The method stays the same. The speed changes everything.