Parallel Consultation with the Internal Seven
In the previous chapter, we noted that the "7" refers to seven internal roles.
Knowing that there are seven roles is one thing. Without a clear picture of what each one actually does, it is hard to get a real feel for how the system works. This chapter covers what each of the seven roles is responsible for, and why we ask all of them for input at the same time.
The Seven Roles and What Each One Looks At
The seven internal subagents (AI agents working inside the organization) each cover a distinct area of expertise. Their domains are designed not to overlap.
Tech Lead handles technical feasibility. The questions here are: "Will this design actually work?", "Where are the bottlenecks (the points most likely to cause a slowdown or failure)?", and "Are there problems that will be difficult to fix later?" Tech Lead evaluates whether an implementation is realistic.
COO looks at overall operations (the day-to-day flow of how things run). The focus is on whether a given approach will hold up in practice, and whether the schedule and workload are reasonable.
QA (Quality Assurance) evaluates the quality of what we produce. For an article, this means checking readability, accuracy, and logical consistency. For a system, it means verifying stable behavior and catching anything unexpected.
Brand Voice maintains consistency in how things are expressed. It asks: "Does this match the tone of the series?", "Are there any words we should not use?", and "Will this actually reach the target reader?"
Task Dispatcher manages the flow of tasks. It checks whether each action is going to the right role, and whether there are any gaps in the dependencies (situations where one task must finish before another can begin).
Researcher handles information verification and the accuracy of our research. It asks: "Is the information behind this decision correct?", and "Are there things we still need to look into?"
Content Director evaluates from a content strategy perspective. It checks whether an article fits the overall direction of the series, whether it meets reader expectations, and whether it drifts from our stated approach.
What "Parallel" Means
When we ask these seven roles for input, we do not pass the work from one to the next in sequence. We send it to all of them at the same time. That is what we call parallel consultation.
Here is a concrete example. Say we want feedback on a draft article.
In a sequential (one-at-a-time) setup, the draft goes to Tech Lead first, then to COO after Tech Lead finishes, then to QA after COO finishes. Each role sees the opinions of the previous ones before forming their own.
In a parallel setup, the same draft goes to everyone at once. Tech Lead, Brand Voice, Researcher — all of them look at only their own area of expertise, without reading what anyone else has said yet.
What Changes When You Run in Parallel
Not reading the previous opinion means not being pulled by it.
In a sequential setup, if the first opinion is strong, the roles that come after tend to be influenced by it. A thought like "Tech Lead said there are no technical problems, so there probably aren't any" can quietly slip into how quality or tone are evaluated. This is a form of conformity bias (the tendency to go along with what others have already said).
In a parallel setup, each role's input comes purely from its own area. Even if Tech Lead sees no problems, QA can still flag something from a different angle. Even if Brand Voice approves, Researcher can still point out that something needs more supporting evidence.
Because the perspectives stay uncontaminated by each other, there is a better chance of catching blind spots (things that might otherwise be missed).
Summary
Each of the seven roles covers a different area of expertise, and their domains are designed not to overlap. Together they address seven perspectives: technical, operational, quality, expression, task flow, research, and strategy.
By having all seven share their input simultaneously and independently, we keep each perspective clean — free from the pull of what others have said first.