Human Architecture™
Design the work before you train the tool.
Human Architecture is the discipline of designing the outcomes, roles, decision rights, leadership behaviors, guardrails, and trust conditions that help people succeed through change. The H.U.M.A.N. framework is Merdeka's specialized model for AI-enabled work.
Definition
Human Architecture is the system around the strategy.
It is how leaders intentionally design what people are being asked to experience, decide, learn, own, and do differently.
The H.U.M.A.N. framework
Five decisions. One discipline: keep the human at the center of every one.
Human Outcome First
What human experience, capability, relationship, or decision should improve?
Use Case Discipline
What friction does this solve, what human value does it protect, and what risk could appear?
Manager Reinforcement
What will managers model, coach, inspect, and normalize?
Accountability + Guardrails
Where do we require human approval, explanation, escalation, or override?
New Work Design
What should be automated, assisted, amplified, or anchored in human judgment?
Work design
The four work-design decisions
Automate
Repetitive, low-risk, rules-based tasks.
Assist
Drafting, summarizing, organizing, preparing, translating, and analyzing—with human review.
Amplify
Decision support, coaching, personalization, insight, and scenario planning.
Anchor
Trust, empathy, ethics, care, teaching, judgment, relationship, and leadership.
In practice
What AI can carry
FAQs, reminders, checklists, summaries, drafted follow-up, preparation, and pattern detection.
Anchored in people
What humans stay responsible for
Complex decisions, emotional conversations, exceptions, equity-sensitive cases, relationship repair, and ethical judgment.
Human Architecture Pulse Check
Ten questions. One honest read.
Rate each statement: 1 = not true yet, 5 = consistently true.
1.We can clearly name the human experience or capability this change should improve.
2.We measure whether the human outcome improved—not only whether the tool or process was used.
3.Each use case solves a specific friction point and has a defined owner.
4.We understand the human value and risk associated with each use case.
5.Managers know what to model, coach, inspect, and normalize.
6.Managers have a repeatable cadence for discussing friction, trust, and behavior.
7.Human approval, explanation, escalation, and override rules are visible.
8.Ownership for accuracy, bias, privacy, and impact is clear.
9.People know what they should do differently on Monday morning.
10.We have stopped, redesigned, or reassigned work—not simply added AI or change on top.
0/10 answered
This is a directional reflection tool, not a validated diagnostic. Responses stay in your browser and are not stored.
Your AI strategy will not be remembered by the tools you bought. It will be remembered by whether your people felt replaced, rescued, or reimagined.