For a long time I used AI assistants the way most people do: open a chat, ask a question, get an answer, close the tab. This was cool, and it provided value. I could do things with less manual research. Neat.
Then I upgraded my toolkit a bit. I started using tools like ChatGPT and Claude to write drafts of SOPs for me. To start turning my brain dumps (often spoken into the phone while driving to school pickups) into organized, shareable artifacts for managing the team. This was novel, and I felt like it was helping me do my work better, but it began to feel like it was managing me.
Then I started assigning roles, setting conversation parameters, and making it work for me. That was the "oh shit" moment where I knew I was barely scratching the surface of what I could be doing.
Now, every project has a defined scope, every conversation topic has a wakeup word, a Notion section it owns, and a brief to read before responding, and to update after we pause. My finance advisor Bruce reads the MRR model and the P&L structure before I ask a budget question. The marketing advisor reads the campaign data and GTM plan. The devops advisor reads the infrastructure doc.
The job title is just context. But the context changes everything. When an advisor has ownership over a specific domain and is expected to update it, the outputs are materially different from a generic chat response. Tighter, more consistent, more useful for actual decisions.
The other thing that changed: I stop repeating myself. The advisor reads the Notion page. Everything I've decided is already in there. The conversation picks up where the last one left off.
Not magic. Not a thinking, self-aware brain with personality. Just structure, rules, and context applied to a tool that elevates what it can do.