30-Day Tuning Window

Voice Constitution Refinement Process

Profit for Contractors

What This Is

The Voice Constitution is your unique system prompt—it makes sure every AI-generated piece sounds like you. But no voice model is perfect on day one. This 30-day window is your structured process for refining it based on real usage and your feedback.

Your time commitment: 15 minutes per week. That's it. You review content as you normally would and flag what doesn't sound right.

Section 1: How the Tuning Window Works

30-Day Timeline Overview

The tuning window runs for 30 days after your Voice Constitution is delivered. During this period, your team uses AI prompts that generate content, and you provide feedback that makes the system smarter.

Here's what happens:

Week 1: Catch obvious voice misses (wrong words, wrong tone, wrong structure)
Week 2: Identify patterns and fix systematic issues
Week 3: Fine-tune nuance and refine prompts for specific content types
Week 4: Lock in the voice and hand over to your team with confidence

Who Participates and How

Austin Maginnis (You)

Source of truth. You review content and flag what doesn't sound right. Your feedback is the north star.

Dayani (Marketing Director)

Primary user. Runs the Master Prompts daily, generates content, collects your feedback, submits it to Kitala.

A. O'Neil (Content Strategist)

Secondary user. Uses prompts for planning and strategy content. Flags issues and contributes feedback.

Dave Ryan / Kitala

Updates the Voice Constitution and Master Prompts based on your feedback. Keeps the system improving.

Expected Time Commitment

For you: 15 minutes per week maximum.

You don't need to review everything. Just flag pieces that don't sound right when you see them. Dayani will collect feedback and send it to Kitala once a week.

What Gets Tuned

Voice Constitution: Your tone, vocabulary, sentence patterns, off-limits words
Master System Prompts: The instructions that shape how the AI writes
Content Engine: Template structure, topic relevance, repurposing quality
Channel Formatters: Caption length, CTA style, how it fits each platform

Section 2: The Feedback System

Three types of feedback matter. The first two are obvious. The third one—what's exactly right—is equally important to capture.

This Doesn't Sound Like Me

The AI used the wrong words, wrong tone, or wrong structure. Something feels off.

This is Exactly Right

The AI nailed it. This sounds like you. Tag it so Kitala knows what's working.

I'd Say It Like This

You provide the exact phrasing you'd use instead. This shows Kitala your preferred language patterns.

How to Submit Feedback

Option 1: Screenshot + Note (Easiest)

When you see content that needs feedback:

Take a screenshot of the piece
Add a quick note: "This doesn't sound like me because..." or "This is great because..."
Send it to Dayani

Option 2: Use the Feedback Form (In This Document)

Print or screenshot the feedback form at the end of this document. Fill it out and send to Dayani.

Option 3: Voice Memo (For Detailed Feedback)

If you have detailed thoughts, record a voice memo on your phone saying what's off and how you'd say it instead. Dayani will transcribe it and submit to Kitala.

Submission Schedule: Dayani collects feedback throughout the week and submits all notes to Kitala every Friday. Kitala reviews and implements changes by the following Monday.

Section 3: Week-by-Week Plan

Week 1
Days 1–7

First Contact: Catching Big Misses

Your team uses Master Prompts for their first real content batch
You review and flag obvious voice misses (wrong words, wrong tone, wrong structure)
Dayani collects feedback and submits to Kitala on Friday
Expected feedback: 5–10 corrections

Your role: Flag what doesn't sound right. Don't overthink it.

Week 2
Days 8–14

Pattern Recognition: Finding Systematic Issues

Your team has enough output to see patterns now
You spot systematic issues (e.g., "AI keeps using passive voice" or "too many questions at the end")
Kitala revises Voice Constitution based on Week 1 feedback
Mid-point review call with you, Dayani, and Kitala (15 minutes)

Your role: Identify patterns. The call is a quick check-in to align.

Week 3
Days 15–21

Fine Tuning: Catching Nuance

Updated Voice Constitution is now in use
You catch subtle misses that only show up with volume
Kitala refines Master Prompts for specific content types (emails, posts, scripts, etc.)
Your team starts self-correcting (good sign—the system is working)

Your role: Keep flagging edge cases. Less feedback needed now.

Week 4
Days 22–30

Lock and Load: Ready to Scale

Voice should be 90%+ accurate now
You review edge cases and special content types (sales copy, webinar promos, DMs)
Final Voice Constitution update
Handoff meeting: your team is self-sufficient

Your role: Approve the final system. You're done tuning.

Section 4: Feedback Form Template

Print this form or take a screenshot when you have feedback to submit. Dayani will forward all completed forms to Kitala on Friday.

Voice Constitution Feedback Form

Date:

Your Name:

Content Type:

What the AI Generated (paste or describe):

What's Wrong (or What's Right):

How You'd Say It Instead:

Priority:

Which Week?

Additional Notes:

💡 Tip: Print this form or screenshot it. Fill it out by hand or digitally and send to Dayani. She'll forward all feedback to Kitala by Friday.

Section 5: Success Criteria

How do we know the tuning window worked? Here's what success looks like:

90%+ Accuracy — You review content and say "yeah, that's me" more than 9 out of 10 times.

Reduced Oversight — Dayani can generate content without you reviewing every single piece.

Light Editing Only — Content needs minor tweaks, not rewrites from scratch.

Faster Production — Content production speed doubles without quality dropping.

Team Confidence — Dayani and A. O'Neil stop second-guessing the AI and trust the system.

Ready to Scale — Your team is self-sufficient. You're not needed for every output anymore.

The Two-Question Quality Check: Before any content goes live after Day 30, ask: (1) Does this read as AI-written? If yes, it fails. (2) Would I say this out loud with no prep? If no, simplify.

Section 6: After Day 30

What Ongoing Maintenance Looks Like

After the 30-day tuning window, your Voice Constitution doesn't stay static. Your message will evolve. Your audience will change. New platforms will emerge.

Quarterly Voice Refresh: Every three months, review how the AI is performing. Flag any drift. Kitala updates the Voice Constitution to keep it aligned with how you're speaking now.

When Your Voice Will Naturally Evolve

You're reaching a new audience (different contractor level, different pain point)
Your message is shifting (new offer, new positioning)
Your personal voice has evolved (new stories, new framework)
Your business model is changing (pricing, packages, business structure)

When to Call Kitala Back

Don't wait for quarterly refresh if:

You're launching a major new offer or positioning change
You're rebranding or shifting your audience significantly
Your voice feels noticeably different than the AI's output
You want to add new content types or channels your Voice Constitution doesn't cover

These updates are straightforward—often a single conversation and a Voice Constitution refresh. No need for another full 30-day tuning window.

You're in Control

The Voice Constitution is a living document that gets smarter based on your feedback. The 30-day tuning window is designed to be minimal lift for maximum accuracy.

Your job is simple: When you see content that doesn't sound right, flag it. Dayani handles the rest. Kitala updates the system. By Day 30, your AI sounds like you.

Questions? Reach out to Dayani (primary) or Dave Ryan / Kitala (if needed).