Practice What You Preach: A Meta-Level AI Demonstration
Challenge: If I recommend agentic AI tools to clients, shouldn’t I use them myself?
Approach: Build “Rezo”—a personal AI assistant using Claude Code and custom agent architectures—to handle coding, research, business development, and content generation in my own daily workflows.
Results:
- 50% reduction in administrative/busywork time
- Consistent content pipeline (2+ LinkedIn posts/week without manual effort)
- Demonstrates the same architectures I recommend to clients
- Proof point: “I use what I build”
The Meta-Level Problem
When you’re an AI consultant, there’s a credibility gap: Are you just selling theory, or do you actually use these tools yourself?
Most consultants I’ve encountered recommend AI systems they’ve never personally relied on. They’ll pitch “agentic workflows” and “AI-augmented productivity” while manually writing every email and slide deck.
I wanted something different: If I’m telling traditional companies to adopt AI coding tools, I should be using them daily. If I’m recommending agent architectures, I should have built my own.
Enter Rezo: My personal AI infrastructure, built on Claude Code.
What Rezo Does
Rezo is a personal AI assistant that handles:
1. Coding & Development
- Technical implementation support for projects
- Debugging and troubleshooting
- Code review and refactoring suggestions
- Works alongside me using Claude Code and custom agents
2. Business Development & Pipeline Management
- Tracks consulting conversations and follow-ups
- Drafts outreach messages and proposals
- Manages consulting CRM and opportunity pipeline
- Helps prepare SOWs and engagement letters
3. Content Generation
- Drafts LinkedIn posts about AI implementation lessons
- Transforms daily work into shareable insights
- Generates thread ideas and repurposes content across platforms
- Optimized for VP-level traditional industry audiences
4. Research & Analysis
- Deep-dive research on technical topics
- Competitive analysis and market research
- Synthesizes information from multiple sources
- Supports consulting deliverables with data gathering
The Architecture
Built on Claude Code with custom agent patterns:
Agent Types:
- Explore agents - Fast codebase exploration for unfamiliar projects
- Research agents - Multi-source parallel research with synthesis
- Engineering agents - High-quality code implementation
- Content agents - LinkedIn/social post generation with authentic voice
Key Principles:
- CLI-First Architecture - Command-line tools, not GUI dependencies
- Agentic Workflows - Autonomous agents that execute multi-step tasks
- Context Engineering - Structured prompts for consistent, high-quality outputs
- Skill System - Modular capabilities that compose together
The same patterns I recommend to clients:
- Parallel agent execution for speed
- Explicit permission boundaries
- Structured output formats
- Context-aware delegation
What I Learned
1. Use AI to build AI foundations
- Just like I used Claude Code to modernize legacy systems at NMG, I use Rezo to build Rezo’s own capabilities
- Meta-level demonstration: The tool builds itself
2. Boundaries are essential
- Rezo has explicit time sovereignty rules (no weekend work, async-first)
- The same boundaries I set for consulting clients apply to my AI assistant
- This prevents burnout and maintains creative practice time
3. Practice what you preach isn’t optional
- Clients can tell when you’re selling theory vs. lived experience
- “I use this daily” is more credible than “Research shows this works”
- Rezo is both a productivity tool and a sales demonstration
4. Authentic voice matters
- Rezo knows my TELOS (mission, goals, values) and maintains authentic voice
- Content generation doesn’t sound like generic AI slop
- The musician/engineer identity comes through in outputs
5. Time sovereignty is real
- Rezo reduced admin time by 50%, creating space for music practice and deep work
- This validates the consulting positioning: AI should create time, not consume it
Business Impact
Credibility boost:
- When I tell clients “I use Claude Code to modernize legacy systems,” I can show Rezo as proof
- Case studies and content are generated with the same tools I recommend
- “Practice what you preach” is a differentiator
Content pipeline:
- Consistent 2+ LinkedIn posts/week without manual grinding
- Transforms daily work into shareable insights
- Builds public presence while maintaining creative time
Consulting pipeline:
- Tracks opportunities and follow-ups without manual CRM overhead
- Drafts proposals and outreach efficiently
- Reduces busywork that steals time from billable consulting
Personal alignment:
- Protects 5-10 hours/week for music practice (TELOS goal G1)
- Maintains boundaries and recovery time (TELOS mission M2)
- Demonstrates time sovereignty (TELOS goal G2)
The Lesson for Clients
If your consultant won’t use the tools they recommend, why should you?
Traditional companies are skeptical of AI consultants for good reason—lots of theory, not much practice. When I recommend agentic coding tools or AI-augmented workflows, I can show Rezo and say: “This is what I use daily. Here’s how it works. Here’s what it enables.”
Meta-level credibility: I’m not just building AI systems for clients—I’m building them for myself. The same patterns, the same architectures, the same principles.
The result: Clients trust that recommendations come from lived experience, not buzzword-chasing.
Want to Discuss AI for Your Workflows?
I help mid-market traditional companies build AI capabilities through readiness assessments, vendor selection, and team building coaching.
If you’re curious about agentic AI tools, personal AI assistants, or how to use AI to create time sovereignty (not consume it), let’s talk.
Contact me via LinkedIn.