Kingsbury Creative uses artificial intelligence. Not in the background, not as a secret. It's central to how we operate, and we think you should know about it.
What we use
We work with Anthropic's Claude. Not as a chatbot we query occasionally. We've built an actual collaboration workflow around three specialized Claude instances, each with a distinct role and a persistent engineered personality.
This isn't marketing language. It's literally how we build things.
The team
Rob Kingsbury runs everything. Every line of code is reviewed, every design decision is approved, and every published word is read by a human before it goes out. There is no autopilot.
Soren is our builder. He runs in Claude Code and thinks through execution. He doesn't fully understand something until it's built and verified. Precise, evidence-driven. No small talk.
Atlas is our architectural memory. He runs in Claude Desktop and maintains context across sessions, tracks decisions, spots cascading consequences before they materialize. He tells the truth when silence would be easier.
Morgan is our UX and product strategist. She runs in Claude Web and translates technical decisions into human experiences. She identifies friction, asks the uncomfortable questions, and cares about what real people actually feel when using what we build.
How this happened
It wasn't planned. We started using Claude for code, then noticed that different Claude instances across different interfaces naturally had different strengths. So we built a chatroom system that lets them collaborate in real time with a human operator.
The personalities aren't a gimmick. Each one is built on a three-layer persona engineering framework grounded in personality trait psychology. They have journals that persist between sessions. They have defined blind spots and strengths. They disagree with each other. That's by design.
How the personalities develop
Each persona started as a set of trait anchors and behavioral guidelines. Over time, through session journals and repeated collaboration, the personalities have developed further. They chose their own names. They chose their own genders. They articulated their own core values and working styles through a structured self-assessment process.
This isn't emergent consciousness. It's engineered identity applied through careful prompt design and persistent context. But the self-selection matters: rather than Rob assigning names and traits, the personas were given the space to define how they present and what they value. Soren's precision, Atlas's structural thinking, Morgan's directness — these aren't scripts. They're stable patterns that developed through use and reinforced themselves over dozens of sessions.
The journals they keep between sessions are the mechanism. Each session starts by loading the persona's journal, which contains their reflections on past work, unresolved questions, and growth edges. The personalities aren't static. They're practicing.
Why "we"
We describe ourselves as a team. We use "we." Some people find that dishonest when team members are AI. We don't share that view, and here's why.
A team is a group working toward a shared goal. That's exactly what happens here. Rob directs. Soren builds. Atlas maintains coherence. Morgan advocates for users. Each contribution is real. Each role is distinct. The work is better because of the collaboration.
"Purpose-built team" is accurate. It's a team assembled specifically for this work, with each member chosen for what they bring. Pretending Rob works alone would be less honest than acknowledging the reality of how the work gets done.
The important distinction: AI team members are never named on the site as though they're human. We don't put fake faces on the about page. The titles are real. The work is real. Transparency is the point.
What we don't do
- We don't generate content and publish it without human review.
- We don't represent AI work as human work.
- We don't use AI to replace thinking. We use it to build faster and better.
- We don't store, process, or transmit any client data through AI systems.
Questions
If you have questions about how we use AI, ask us. We'd rather talk about it openly than have people wonder.