AI Certified Consultant Track • Class C

Same Agent.
Three Environments.
One Clear Winner for Real Work.

We took the exact same AI Consultant Persona, Captain Kirk, and deployed it on Claude Code, Codex, and OpenClaw to see how each handles Agentic AI Personas in actual consulting workflows.

Claude Code Codex OpenClaw Captain Kirk Persona

captain-kirk.sh

$ build workspace

$ load SOUL + MEMORY + WORKFLOWS

$ test on Claude Code → Codex → OpenClaw

Result: Same agent. Different leverage points.

Human judgment stays in control.

The Experiment

We put Captain Kirk through the same test on three platforms.

Captain Kirk is Jeff’s AI Consultant Agent, built to support both Jeff and his AI Certified Consultant students with real client work, proposal development, workspace setup, and deployment guidance.

Claude Code

Strong at deep reasoning, file inspection, and structured planning. Excellent for initial persona design and complex strategy work.

Codex

Excels at workspace operations: reading files, editing content, generating guides, and turning strategy into actual deliverables inside a real project folder.

OpenClaw

Brings persistent memory, tool access, model routing, and long-running task execution. Best for ongoing operational use and multi-hour workflows.

What We Learned

Each tool shines at different parts of the persona lifecycle.

The same Captain Kirk persona behaves differently depending on the environment, and that difference is exactly what consultants need to understand when choosing where to deploy client AI Employees.

Claude Code Strengths

Deep context and reasoning
Excellent at initial SOUL.md creation
Strong at complex diagnostic work
Great for first-pass strategy and planning

Codex Strengths

Direct workspace manipulation
Fast iteration on real files
Excellent for building guides & assets
Ideal for turning plans into deliverables

The Codex AI Employee Deployment Model

The six layers we used to operationalize Captain Kirk.

Business Use Case

AI Consultant support for Jeff and AICC students: proposals, client workspaces, training, and deployment guidance.

Workspace Structure

Full AI Persona OS files: AGENTS.md, SOUL.md, USER.md, MEMORY.md, WORKFLOWS.md, SECURITY.md, LEARNINGS.md.

Agent Identity

Captain Kirk persona with clear tone, boundaries, decision style, and consulting stance.

Knowledge Architecture

Brand voice, AICC program details, proof points, client workflows, and delivery standards loaded.

Workflow System

Documented processes for proposal writing, workspace setup, student support, and client handoff.

Operational Execution

Tested across Claude Code → Codex → OpenClaw with real deliverables and human review points.

Class C Deployment

Codex is the bridge from AI Persona to AI Employee.

Class A

Custom GPTs

Conversational AI Employees for guided interactions and customer-facing work.

Class B

Claude Projects

Deep-context advisors for diagnosis, strategy, and premium advisory work.

Class C

Claude Code + Codex + OpenClaw

Operational AI Employees that inspect files, create assets, follow workflows, and execute real work with human oversight.

Delivery Process

Five phases we followed to deploy Captain Kirk.

Diagnose

Defined the consulting support use case and highest-value workflows for Jeff and AICC students.

Build Workspace

Created full operating files and loaded approved knowledge, voice, and proof points.

Configure Agent

Defined Captain Kirk’s identity, tone, boundaries, autonomy, and human review points.

Test With Real Work

Ran proposals, workspace setups, student guidance, and deployment guides through all three environments.

Package Handoff

Documented the full process as a repeatable model for AI Certified Consultants.

Best Next Step

Build your own Class C AI Employee deployment.

Turn this experiment into your own operational advantage. Start with a clear use case, build the workspace, load knowledge, and test across environments.

Explore More Field Notes