Structure matters more than prompts
Much of the public conversation around AI revolves around prompts. Ask a better question and you'll get a better answer.
Russell believes the challenge runs deeper than that.
Early in his own experiments, he found that long conversations with AI often started strong but gradually drifted away from the original objective. Requirements changed, unnecessary features appeared, and the final output didn't always reflect what he had intended to build.
As he explained during the interview:
I needed something to have like a guardrail for my own safety.
That idea of guardrails became the foundation for what he calls the "three-man team," a workflow designed to bring more structure and accountability to AI-assisted projects.
Rather than treating AI as a single assistant, Russell treats it as a team of specialists working together toward a shared outcome.
Meet the three-man team
The framework mirrors a process many agencies already follow.
At the top is an architect who acts as a strategist and project lead. This role helps define goals, clarify requirements, and document exactly what needs to be built before any work begins.
Once the scope is established, the builder takes over. The builder's responsibility is creating the deliverable itself, whether that's a WordPress plugin, a marketing asset, or another project entirely.
Finally, a reviewer compares the finished work against the original requirements to verify that everything aligns before the project moves forward.
Russell described the workflow this way:
My architect actually writes a file and says, this is what we're building. Then my builder reads that file and builds everything. Then my reviewer looks at everything that was built and says, this is what the architect wanted, this is what the builder built, do they meet?
What's interesting is that the system isn't limited to development projects. Russell has created versions of the three-man team for marketing, content creation, support, and other business functions. The individual roles change depending on the task, but the underlying philosophy remains the same: define responsibilities, document expectations, and create a process for reviewing the results.
For agency owners, the idea may feel familiar because it closely resembles the way successful teams already operate. Clients speak with strategists. Strategists create plans. Specialists execute. Reviewers check the work before delivery.
The difference is that Russell has adapted that workflow for AI.
The structure itself solves one problem: keeping projects organized. As Russell continued refining the system, he discovered another challenge. Even with clearly defined roles, the quality of the output often depended on how each agent approached decisions and interpreted the information it received.
Teaching AI how to think
Another layer of Russell's approach involves assigning each agent a distinct persona.
His architect isn't simply labeled "architect." Instead, the role is modeled after experienced WordPress professionals he has worked with over the years. The persona reflects someone who has seen projects succeed, fail, and evolve, and who understands how today's decisions affect future outcomes.
The purpose is to give the AI a clearer frame of reference, helping it make decisions within a more defined set of expectations.
Russell takes the same approach in other areas of his business. Marketing agents are shaped by people he has worked with in marketing roles. Support agents are modeled after experienced support professionals. In some cases, he even creates agents based on his own approach to troubleshooting and problem solving.
The concept highlights something many businesses are beginning to discover about AI. The more context you provide, the more useful the output tends to become. That context doesn't always need to come in the form of instructions. Sometimes it comes from clearly defining the perspective from which decisions should be made.
The role AI cannot replace
For all of his enthusiasm about AI, Russell repeatedly returned to one point throughout the conversation: someone still has to own the outcome.
AI can generate code, recommend solutions, and automate portions of a workflow, but it doesn't eliminate the need for oversight.
As Russell put it:
The buck still stops with me.
That means testing code before release. It means reviewing recommendations before implementing them. It means verifying that the solution solves the intended problem rather than assuming the AI got everything right.
This perspective feels particularly relevant as more businesses experiment with AI-powered workflows. The most effective implementations are rarely the ones that remove humans from the process entirely. Instead, they're the ones that allow people to focus their attention where it matters most.
AI can help accelerate the work, but accountability remains a human responsibility.
That balance between automation and oversight becomes even more important as AI capabilities move closer to the tools people use every day. During the conversation, Russell pointed to WordPress 7.0 as an example of how quickly that shift is happening.
What WordPress 7.0 changes
The conversation naturally turned toward WordPress 7.0 and the growing role AI is expected to play within the platform itself.
For years, AI-powered workflows largely existed outside of WordPress. Users relied on separate tools and services to access AI capabilities. WordPress 7.0 begins bringing some of those experiences closer to where people already work.
Russell pointed to new connector functionality, the Abilities API, and the broader effort to establish standards around how AI integrates with WordPress moving forward.
What excites him most is how AI is becoming part of the broader WordPress experience rather than remaining a collection of disconnected external tools.
He described it as another leveling moment for the ecosystem. Similar to previous milestones like custom post types, users now have an opportunity to learn and experiment together as new capabilities emerge.
One example he highlighted was AI-powered title generation. Having AI analyze a completed draft and suggest headline options can help remove friction from the content creation process while leaving creative decisions in the hands of the author.
Throughout the discussion, Russell framed AI as a tool for reducing repetitive work and expanding what individuals can accomplish on their own. WordPress 7.0 represents another step in that direction by bringing more of those capabilities directly into the dashboard.
Building systems that scale
The technology behind Russell's three-man team is interesting, but the underlying process is what makes it effective.
The examples he shared throughout the conversation involved tasks that consume time without necessarily requiring deep strategic thinking: organizing information, generating reports, reviewing data, or moving work from one format to another. Those activities are necessary, but they're rarely where agencies create the most value for clients.
By reducing the time spent on repetitive work, AI creates more room for strategy, relationship building, problem solving, and creative thinking. The expertise agencies have spent years developing remains important. AI simply changes how some of the supporting work gets done.
For agencies looking to put these ideas into practice, the first step isn't building a team of AI agents. It's understanding your existing workflows.
Start by identifying a task that is repetitive, time consuming, or difficult to scale. Document how that work currently gets done, including the decisions being made along the way and who is responsible for making them. Once the process is clear, look for opportunities where AI can support specific steps without removing the review and accountability that keep quality high.
That's ultimately the lesson behind Russell's three-man team. The framework itself is less important than the thinking that shaped it. Clearly defined responsibilities, documented expectations, and deliberate review processes help people produce better work. AI can accelerate those systems, but it works best when there's already a clear path to follow.
As AI capabilities continue to expand inside WordPress and beyond, the businesses that benefit most may not be the ones chasing every new tool. They'll be the ones building thoughtful, repeatable systems that allow those tools to support the work they're already doing well.
.webp)
