VINCH
E01
VINCH / VINCE LIEM / ENGINEERING

Engineering moves faster when people own the change, the judgment, and the result.

Purpose

AI can accelerate implementation; it does not own the result. I turn intent into readable behavior, keep architecture and review under human authority, and use agents inside a traceable path from issue to verified change.

Concrete underpass with an illuminated staircase
E02 / Engineering control loop

A deliberate control loop keeps change legible.

AI can propose implementation at speed. The engineering system keeps behavior explicit, change reviewable, and responsibility with the people who own the product.

Engineering control loop. Issue, readable behavior, BDD and Gherkin, agent implementation, pull request, review and checks, merge, and feedback. Feedback returns to behavior and architecture.
  1. 01 / Human intent

    Issue

    Capture the real problem, opportunity, and operating context.

  2. 02 / Human behavior

    Readable behavior

    Describe expected behavior in language the people involved can inspect.

  3. 03 / Human specification

    BDD / Gherkin

    Turn intent into concrete Given, When, Then scenarios.

  4. 04 / Agent execution

    Agent implementation

    Generate a change against explicit behavior and architectural direction.

  5. 05 / Human review

    Pull request

    Expose context, rationale, differences, and trade-offs for review.

  6. 06 / Human judgment

    Review and checks

    Combine human judgment, collaboration, automated checks, and verification.

  7. 07 / Human acceptance

    Merge

    Integrate the accepted change while behavior remains the source of truth.

  8. 08 / Product learning

    Feedback

    Use product learning and code feedback to refine behavior and architecture.

Feedback returns to behavior and architecture.

E03 / Human authority

Generation is leverage. Ownership stays human.

Human authority model. Behavior, architecture, review, and verification remain within a human authority boundary while agent execution enters through explicit gates.
  1. 01

    Behavior before implementation

    Decide what should happen before deciding how the code should change.

  2. 02

    Architectural judgment

    People own boundaries, structure, risks, and consequential trade-offs.

  3. 03

    Reviewable change

    Small, explainable changes compound into systems that remain safe to evolve.

  4. 04

    Independent verification

    Tests and checks support acceptance; they do not replace accountable judgment.

Principles / operating constraints

The rules that keep leverage accountable.

  1. 01

    Legibility

    If people cannot read it, they cannot responsibly own it.

  2. 02

    Explicit behavior

    Intent is inspectable before implementation begins.

  3. 03

    Collaboration

    Humans and AI work together; authority stays with humans.

  4. 04

    Reviewability

    Every change remains traceable, testable, and discussable.

  5. 05

    Teachability

    The method can be learned, challenged, and improved.

Continue through the operating modelSystems