QUICK START GUIDE
From zero to PAMO in 90 days
═══════════════════════════════════════════════════════════════
WHO THIS IS FOR
═══════════════════════════════════════════════════════════════
You are a Principal (founder, owner, CEO with skin in the game).
You've read "To the Principal" and you're convinced.
You want to act. Here's how.
═══════════════════════════════════════════════════════════════
DAY 1-7: MAP
═══════════════════════════════════════════════════════════════
Task: Draw your real org chart. Not the HR one.
The one that answers: "Who coordinates whom?"
For each person, mark:
[M] = Maker (directly produces value customers pay for)
[C] = Coordinator (ensures others do their jobs)
[P] = Principal (you, or sub-Principals with real authority)
Count the [C] nodes. That is your bureaucratic load.
Multiply by average [C] salary. That is your annual
cancer-feeding cost.
Deliverable: One page. Three columns.
Name | Role (M/C/P) | Primary function
═══════════════════════════════════════════════════════════════
DAY 7-14: PICK ONE FUNCTION
═══════════════════════════════════════════════════════════════
From your [C] list, pick the ONE coordination function that:
- Is most repetitive
- Involves the most information routing
- Has the least judgment requirement
- Annoys Makers the most
Common first picks:
- Weekly status reporting
- Task assignment and tracking
- Meeting scheduling and agenda distribution
- Onboarding information delivery
- Project progress dashboards
Do NOT start with:
- Performance evaluation (too political)
- Budget allocation (too sensitive)
- Strategic planning (too complex)
Start boring. Start small. Prove the concept.
═══════════════════════════════════════════════════════════════
DAY 14-30: DEPLOY FIRST SAGENT
═══════════════════════════════════════════════════════════════
Build or configure an AI agent that performs that ONE function:
- Receives inputs (from Makers or systems)
- Routes information (to relevant parties)
- Tracks completion (deadlines, status)
- Reports transparently (no narrative, just data)
- Logs everything (immutable record)
Technology options (2026):
- Claude/GPT with structured workflows
- Custom agent with tool-calling
- NiceGUI dashboard + LLM backend
- Any stack that produces: route + track + log
The tool doesn't matter. The principle matters:
The coordinator has no self-interest.
═══════════════════════════════════════════════════════════════
DAY 30-60: MEASURE AND COMPARE
═══════════════════════════════════════════════════════════════
After 30 days of the Sagent running one function:
Measure:
1. Information latency: how fast do you get status now?
(Compare: before = waiting for weekly report from Manager)
2. Information accuracy: is data raw or narrativized?
(Compare: before = Manager's interpretation/spin)
3. Cost: what did that function cost in salary before?
(Compare: Sagent operational cost vs human salary)
4. Maker satisfaction: do Makers feel less overhead?
5. Resistance: who is uncomfortable? Why?
The resistance signal is diagnostic:
- If Makers love it → you're on the right track
- If a [C] person feels threatened → you found cancer
- If no one notices → function was already dead weight
═══════════════════════════════════════════════════════════════
DAY 60-90: EXPAND OR ITERATE
═══════════════════════════════════════════════════════════════
If first Sagent succeeds:
→ Pick second function from [C] list
→ Deploy second Sagent
→ Begin building the QR/QD measurement layer (Ch9, Ch10)
→ Consider OSA role: who will architect your system?
If first Sagent fails:
→ Diagnose: wrong function? Wrong tool? Sabotage?
→ Read Ch28 (full implementation path)
→ Contact author for guidance
═══════════════════════════════════════════════════════════════
THE 90-DAY CHECKPOINT
═══════════════════════════════════════════════════════════════
After 90 days you should have:
☐ Real org chart with M/C/P labels
☐ At least 1 Sagent running a coordination function
☐ Before/after metrics on that function
☐ A list of next functions to convert
☐ An understanding of where resistance lives
☐ A decision: go deeper or get help
═══════════════════════════════════════════════════════════════
WHAT FULL PAMO IMPLEMENTATION LOOKS LIKE
═══════════════════════════════════════════════════════════════
Stage 1 (you are here): Single function Sagent
Stage 2: Multi-function Sagent layer (QR tracking)
Stage 3: QD measurement + transparent value signals
Stage 4: Full Oagent tree + real-time audit chain
Stage 5: Autonomous coordination (humans at legal nodes only)
Most companies stop at Stage 2-3 and already see 3-5x
improvement in coordination efficiency.
Full Stage 5 = the 10,000-person company with 20-person
coordination cost. That's the endgame.
═══════════════════════════════════════════════════════════════
FURTHER READING
═══════════════════════════════════════════════════════════════
Ch4: Why bureaucracy is mathematically inevitable
Ch18: Your company's lifecycle stage
Ch24: Sagent architecture (5 modules)
Ch28: Full Monday Morning implementation path
App C: One-pager for startup founders
→ Contact: contact.html