Organization Structure
Leadership roles become architects of constraints. When AI handles execution, the value of every leadership role shifts to designing the rules, processes, and boundaries that others — and AI — operate within.
Three Architects, Three Domains
An organization needs constraints in three areas. Each has a different owner.
| Role | Architect of… | What they design |
|---|---|---|
| PM (Business Architect) | The product | Feature strategy, business rules, what to build, success criteria |
| Manager (Org Architect) | The operations | Communication rules, tools, team missions, internal processes |
| Technical Architect | The system | Technical constraints, AI agents, code rules, quality guardrails |
PMs are business architects. They translate the company vision into detailed business rules for daily work. What to build, why, what success looks like. These decisions require understanding the market, the customer, and the strategy — AI cannot own them.
Managers are organization architects. They design how the company works internally — communication norms, tools, team boundaries, what each team owns. They create the environment where people and AI work well together.
Technical architects design the system. They define code conventions, build AI agents, set quality guardrails, and create the rules that make AI output consistent and trustable. When something goes wrong, they improve the rules — not review every line.
All three design constraints for others to operate within. Different domains, same function.
Members
Members build within all three sets of rules. Business rules tell them what to build. Organization rules tell them how to work. Technical rules tell them how the system operates.
- Broad knowledge over deep specialization. See
engineering-teamsfor details on the generalist position. - Problem-solving and judgment are the core skills — AI handles execution.
- Members work freely within the safe area that architects defined.
Constraint Authority
Rules must come from a small number of people. One architect per domain per team. If teams are large, sub-architects can handle delegation, but decision authority stays concentrated.
- Members can propose rules and submit them for approval.
- Architects approve, improve, and maintain the rules.
- This is intentionally a bottleneck — constraint design cannot be distributed to everyone without losing consistency.
The Junior Problem
AI removed the execution work juniors used to learn through. Before AI, juniors built intuition through repetition — debugging, refactoring, seeing patterns. Now AI handles most of that.
- Decision-making and problem-solving must be taught earlier and more deliberately.
- Good guardrails create a safer space for juniors — clear rules, defined conventions, automated quality checks.
- Guardrails reduce risk but do not replace learning. Judgment is still human.
AI as Thinking Partner
AI is more than an execution tool. It can research, propose options, surface tradeoffs, and challenge assumptions. Every role uses AI to expand what they can consider.
The decision stays human. AI presents options. The human who approved the work carries the consequences.
Decision Criteria
When structuring an organization: identify who owns constraints in each domain (business, operations, technical). If nobody owns a domain explicitly, constraints will be inconsistent.
When deciding architect count: prefer one per domain per team. Scale with sub-architects only when the team is too large for one person to maintain the rules.
When deciding member specialization: prefer breadth. Deep expertise belongs in the architect role. Members need enough breadth to evaluate AI output across areas.
Anti-patterns
- No clear constraint owner — when nobody owns the rules for a domain, every person and every AI tool makes different decisions.
- Everyone designs constraints — distributing rule-making across all members produces inconsistency. A small number of architects must own this.
- Delegating judgment to AI — AI expands options, humans decide. The moment anyone delegates their judgment to AI, quality breaks.
- Same structure as before AI — if the org structure has not changed since adopting AI, it is likely misaligned. Execution shifted to AI, but the roles around it must shift too.