This post is part of an ongoing series about what we’ve learned from augmenting our team with agents. This series shares hard-won lessons from integrating agents into our team. It’s not theory—it’s transformation, in motion.

AI agents struggle to succeed when you treat them like tools. We know this because we’ve been down this path, and 82% of enterprise led agentic projects are shelved after 12 months. “FastTrack Tool #17” won’t spark enthusiasm and drive usage. But Reese and Casey? They changed the conversation.

Here’s what we’ve learned by doing the work:

1. Personas Build Teammates

When we first started talking about agents, the most common reaction wasn’t excitement—there was an undercurrent of fear and anxiety. People worried they’d be replaced.

But after we introduced Casey as a teammate, things shifted. The conversation became: “How can we help Casey succeed and do more for us?” That reframing worked because Casey felt like a person, a part of the team—not a bot.

2. Onboarding, Not Launching

We learned quickly that you don’t launch a teammate—you onboard them.

For us, that meant treating agents with the same discipline as new hires: communication plans, awareness sessions, training, and buddy systems. Adoption improves the moment you stop thinking “tool release” and start thinking “new colleague.” 

We do regular reviews with our agents’ performance, just like we do with the rest of the team – more frequent right after onboarding (or iterations of new capabilities, just like promotions).  Less frequent as our agents get more experience.

3. Scaling Agents Comes with Responsibility

At first, agents were treated casually—something that was just a test, that could be turned on or off at will. That didn’t work.

Now, our agents are roles on the org chart. Adding or retiring an agent requires a process, because their work has real dependencies. One person’s frustration shouldn’t lead to Winston being deleted on a Friday afternoon any more than it should lead to a human being walked out the door.  We need to put the same thought, coaching, iteration, and decision process into all of our teammates; human or not.

4. Personas Force Clarity

Our early experiments taught us that tool development drifts—overlap, redundancy, and confusion creep in. But personifying agents forces sharper thinking.

When Mona “graduated” from a personal helper to a team-level role, we treated it like a promotion. We clarified scope, aligned expectations, set up an owner (manager), and eliminated overlap. Without that discipline, grassroots innovation can quickly tip into chaos (more on this in a future post).

The Big Lesson

This isn’t theory. This is earned wisdom that we’ve learned by doing.

Giving your agents personas isn’t just branding—it’s adoption strategy, trust-building, team culture, and organizational clarity.

Because when an agent stops being Tool #17 and starts being Theo, your team doesn’t ask “Do we need this?” anymore. They ask “How do we help them thrive?”