GRIFFAIN
Services / AI Systems
← All Services#tiers#fitFind the gap
GRIFFAIN AI · AI SYSTEMS

AI installed into your operation.Not bolted on top.

$2,500 — $12,500·Standard to Retainer · ICM-based
WORKSPACE ARCHITECTUREICM INSTALL · v1.0CLAUDE.mdâ–¸ 01_sales/— CONTEXT.md— leads.md— scripts.mdâ–¸ 02_content/— CONTEXT.md— voice-guide.md— editorial.mdâ–¸ 03_operations/— CONTEXT.md— clients.mdROUTING LOGICClaude reads CLAUDE.md on openNavigates via CONTEXT.md filesStage context orients each taskReference layer is calibrated4 workspaces · 16 stages8 reference files7-day install · 90-min training
chat.openai.com
ChatGPT
Write me a follow-up email for our HVAC client
Dear [Client Name], I hope this email finds you well. I wanted to follow up regarding your recent service... [generic, could be anyone]
No context. No voice. No system.
BEFORE · GENERIC AI
02_content/ → voice-guide.md → email-templates/
02_content/ → voice-guide.md → customer-profiles.md → email-templates/
Hey Mike — following up on the Northside install. Quick note on the timeline you mentioned...The calibrated, specific voice your business actually has.
AFTER · CALIBRATED WORKSPACE

Most AI implementations are calibrated to the wrong picture.

Generic AI tools give generic outputs because they don't know your business. GRIFFAIN AI System builds start with an audit — identifying the corrected ontology before a single folder is created. The workspace that ships is calibrated to how your operation actually works, not how it describes itself.

7 days
Avg. install time
Standard single-workspace. Brief to trained operators, all reference files calibrated.
90 min
Handoff training
Live session. Operators driving the workspace. Written playbook delivered the same day.
5
Build tier workspaces
Sales, content, operations, client success, reference. Full operational coverage.

What ships

Scope adjusts by tier
ICM Audit
Founder and operator conversations. Operational review. Substrate identification — what the business actually is vs. what it says it is. The foundation everything is built on.
Audit precedes every install — the corrected ontology, not the stated one
ICM ARCHITECTUREREFERENCE LAYERWORKSPACE LAYERCLAUDE.md
Workspace Architecture
Folder structure designed to the corrected ontology. CLAUDE.md written at root. Routing logic defined before building begins.
CONTEXT.md Files
Every workspace level. Every stage level. The operator's map — any agent or operator knows where they are and what they're supposed to do.
Reference Layer
Voice guide. Customer profiles. Decision criteria. Operational reference. Calibrated to the audit — not generic templates.
Custom Prompt Library
Task-specific Claude prompt templates built for your actual recurring workflows. Tested against real tasks before handoff.
Quality Testing
Every workspace tested with actual tasks — not theoretical. Routing verified. Operators can navigate without a guide before we leave.
Operator Training
60–90 min live session. Your team driving the workspace while we observe. Q&A on extension patterns and maintenance.
Written Playbook + Video
Operating playbook delivered. Walkthrough video recorded. Git repo transferred. You own it completely — no ongoing dependency.
Git repo transfer
Walkthrough video
Written playbook
7-day support

Engagement tiers

First payment before work begins
Standard
1 week
$2,500–$5,000
50% upfront · 50% on delivery
Single workspace install. One operational area covered end-to-end with reference layer and training.
  • Single workspace (4–6 stages)
  • CLAUDE.md + all CONTEXT.md files
  • 5–10 reference files calibrated
  • Custom prompt library
  • 60-min handoff training
  • Written operating playbook
Start diagnostic →
BuildMost Common
2 weeks
$5,000–$12,500
50% / 25% mid / 25% delivery
Multi-workspace install with deep reference layer. The most common engagement for full operational coverage.
  • 3–6 workspaces (full operation)
  • Deep reference layer (voice, customers, decisions)
  • Architecture calibrated to corrected ontology
  • Custom prompt library
  • 90-min live training session
  • Walkthrough video
  • Written playbook + Git transfer
  • 7-day post-install support
Start diagnostic →
Retainer
Monthly
$1,000–$2,500/mo
Net-0 (1st of month)
Ongoing workspace maintenance as the business evolves. For active workspaces that need monthly calibration.
  • Monthly workspace review call
  • Reference material updates
  • New workspace additions as needed
  • Operator support
  • Extension pattern documentation
  • Monthly performance review
Start diagnostic →

Is this a fit?

This is a fit
You or your team spends significant time on tasks Claude could handle with the right context.
You have knowledge work that needs to be consistent — writing, analysis, customer responses, decision support.
Your operations produce structured outputs repeatedly and each one takes manual effort.
You've used ChatGPT and found it useful but inconsistent — no system behind it.
You want AI in your business but you don't know where to start or how to structure it.
This is not a fit
Pure transactional operations — e-commerce order processing. That's automation, not augmentation.
Heavy physical operations where most of the work isn't knowledge work.
Very small operations where the workspace install costs more than the time it saves.
You want AI to replace workers, not augment them. This isn't that model.
Compliance-bound work where AI augmentation creates specific regulatory risk.
Start Here

The diagnostic call
is 20 minutes.
And it's free.

We audit the operation, design the architecture, build the workspace, and train your team. The diagnostic call is 20 minutes. It's free.

Book the diagnostic call
01
ICM audit
Founder conversations. Operational review. We find the corrected ontology before building.
02
Architecture
Workspace structure designed. CLAUDE.md written. Reviewed before build begins.
03
Build + test
Reference layer calibrated. Every workspace tested with real tasks before training.
04
Train + handoff
90-min live session. Your team driving. Playbook and video delivered.
GRIFFAIN AI · AI Systems · Est. 2026griffain.comDiagnostic-first · No menu