Meet Rosey 🤖, Our AI Teammate

At Yulu, we’ve taken a proactive approach to exploring how AI can support – but never replace – our work. That’s why we built Rosey: our very own AI assistant, named after the beloved robot from The Jetsons.

But Rosey isn’t just any bot.

She’s trained on a secure, closed database of Yulu’s internal knowledge and values – designed to work like a seasoned team member. Her job? To accelerate research, fuel ideas, and support our storytelling – while staying rooted in our voice, ethics, and mission.

We love Rosey. She’s not perfect (she still hallucinates from time to time), but we don’t hold it against her. Because at the end of the day, we don’t publish anything AI-generated without two layers of human editing, oversight, and signoff.

AI can support our process – but it can’t replace the strategy, judgment, integrity, or lived experience that our team brings to the table. All final content must reflect the voice, values, and vision of the human author or client.

Not a Love Letter to Efficiency

Let me be clear: this is not a love letter to AI from a CEO thrilled about cutting labor costs. On the contrary.

I see Rosey – and tools like her – as a way to scale our impact, deepen our creative process, and retain our people. Clients today are expecting AI to be embedded in our work, even if they’re not saying it aloud. They expect us to do more, in less time. Rosey helps us meet that need, without sacrificing heart or quality.

And in our line of work – as a B Corp communications agency – our clients still value the human connection more than anything.

There’s a stat that says human-generated content is three times more likely to be read than AI-generated content. And we’ve seen that firsthand.

Human Still Wins

Recently, our lead writer Temidayo wrote a stunning piece for the Impact Communications Institute’s newsletter, The Agenda (read it here). As part of our internal protocol, we ask writers to disclose what percentage of their work was AI-assisted. In this case? Zero.

Still, for curiosity’s sake, I ran the piece through ChatGPT to see if we could “polish” it. The result? The voice dimmed. It lost nuance. I felt the quality dropped by about 20%.

We took one or two small edits – some light sentence smoothing – but otherwise, we went with Temidayo’s original. The final result? Beautiful, human, memorable.

Building Client Agents With Consent & Care

Rosey was just the beginning.

We recently brought in a respected AI strategist to help us design custom AI agents – not just for Yulu, but for our clients (for those who opt in with full transparency and consent).

Here’s how it works:

• We build a secure, private client agent hosted within a paid enterprise-grade ChatGPT environment with advanced privacy settings.
• On the backend, we train the agent on brand guidelines, messaging strategy, past media, bios, LinkedIn data, and more – the same way we’d onboard a new team member.
• We then use that agent to support our work – for brainstorming, editing, research, and content ideation.

Think of it as adding a junior team member to the account – at no added cost. It stretches each retainer dollar further and adds real value for our clients.

And so far? They’re loving it.

Draft co-written with ChatGPT – human-led authorship by Melissa Orozco (approx. 80%)

Curious about how we approach AI? Read our policy here