System Overview
Pivot Protocols is a behavioral stability model designed to manage volatility during withdrawal, taper attempts, and medication transitions.
The system measures instability through daily behavioral signals, calculates a Volatility Density Index (VDI), and detects destabilization patterns that commonly cause taper attempts to collapse.
The Pivot Containment System (PCS) operationalizes this framework by combining a rule-based scoring engine with a constrained LLM guidance layer that delivers structured stabilization guidance.
The goal is simple: stabilize the system first, then allow reductions to occur inside regulation rather than instability.
Note: The Volatility Self-Check page illustrates the primary user input flow that would power the MVP scoring system.
The following pages provide the conceptual and operational context behind the Pivot Containment System and the Mechanics of Instability.
System Foundations
Measurement Layer
These pages describe how instability is evaluated and quantified before stabilization or reduction occurs.
Stabilization Framework
These pages explain the behavioral containment and stabilization strategies used to restore regulation before attempting reductions.
How To Stabilize Before Reducing
Nervous System Support During Withdrawal
Endogenous Opioid System Support
Sleep Disruption During Withdrawal
Supplements for Sleep During Withdrawal
Reduction Logic
These pages explain how stability markers inform structured dose reduction once regulation has been restored..
Suboxone Taper: How To Reduce Without Destabilizing
Real World Use
These illustrate how the framework applies to real transition scenarios. They also serve as user acquisition entry points.
Self Directed vs Structured Taper
Example Case Article
Kratom - Suboxone Transition: Preventing Precipated Withdrawal
System Entry Points
Operationalizing the Model
Minimum Viable System
Pivot Containment System — MVP Developer Overview
The Pivot Containment System is a mobile-first stability tracking tool designed to reduce volatility during medication transitions and tapering.
It uses a rule-based volatility engine (VDI) to determine system state, then an LLM layer to deliver interactive, constrained guidance—focused on containment, sequencing, and stability restoration.
Executive Summary
Rule engine → state evaluation → interactive guidance.
The core of the system is deterministic: users submit daily stability inputs, the system calculates Volatility Density (VDI), evaluates drift rules, and returns a short directive guidance response.
The LLM is not a decision engine. It is a structured interaction layer that:
explains the score
answers bounded “what-if” questions
provides containment actions based on current state
keeps guidance consistent with phase rules
MVP Goal
Deliver a lightweight, branded system that:
captures daily volatility signals
converts them into a VDI score + category + trend
triggers drift alerts when instability escalates
provides interactive guidance that still stays inside clear parameters
What the MVP Includes
Mobile-friendly branded check-in interface
Simple data store (Airtable or Google Sheets)
VDI scoring logic
Drift rule evaluation
LLM-guidance layer with bounded interaction
Operator alerts when volatility escalates
Minimal operator dashboard view
Guardrails
This system is intentionally constrained.
It is not:
a prescribing tool
a dosing or titration engine
a therapy chatbot
an EHR / medical record system
an open-ended “talk about anything” assistant
The purpose is behavioral containment and sequencing support alongside medical care.
Daily Check-In Inputs
Each check-in captures:
client_id
phase (Containment / Reconditioning / Autonomy)
sleep_hours
urge_intensity (1–10)
interval_pressure (1–10)
emotional_amplitude (1–10)
early_redose (yes/no)
timestamp (auto)
VDI Scoring
Inputs convert into a volatility score that determines today’s state.
Example scoring (adjustable in v1):
Sleep
0–3h = +3
4–5h = +2
6–7h = +1
8h+ = +0
Other drivers
Urge intensity = 0–3
Interval pressure = 0–3
Emotional amplitude = 0–2
Early redose = +2
Category
0–3 = Stable
4–6 = Moderate
7–9 = High
10+ = Escalating
Drift Rules
System flags destabilization patterns and alerts the operator when needed.
Examples:
score worsens 3 consecutive days
score ≥ 10
early redose twice within 7 days
no check-in for 48 hours
LLM Guidance Layer
The LLM is always present in the MVP, but it operates inside strict boundaries.
What the LLM can do
translate system state into a clear directive message
explain why today’s VDI is high/low (score breakdown)
recommend containment actions aligned with the current phase
answer bounded questions about the VDI drivers
run “what-if” scenarios using the rule engine logic (not guessing)
What the LLM cannot do
suggest dose changes or medication instructions
override the rule engine or authorize reductions
drift into therapy-style conversation
invent missing data
Interaction Design (How it stays constrained but feels interactive)
After each check-in, the system returns a guidance message plus structured interaction options:
Buttons / modes
Why is my VDI high today? (explain score drivers)
What should I do today? (containment actions)
What if sleep improves tonight? (bounded scenario)
What if urges spike later? (bounded scenario)
Am I stable enough to reduce? (rule-based yes/no + conditions)
Short interaction rule
max 2–4 turns per mode
LLM may ask 1 clarifying question only if it affects scoring (ex: preventative vs reactive redose)
This creates a modern “LLM feel” without losing structure.
Guidance Output Requirements
Each guidance response must:
be 4–6 sentences
use directive tone
include one containment action
include one objective for the next 24 hours
include an optional warning if drift flag is triggered
avoid clinical/dosing language
High-Level Architecture
CORE DECISION LOGIC
The Pivot Containment System operates using a layered decision model:
VDI → Stabilization State → Pathway → Guidance
VDI (Volatility Density Index)
The system measures instability markers including sleep continuity, dosing interval stability, emotional volatility, and redosing pressure.
Stabilization State
Based on the VDI score, the system identifies the user’s current stabilization state within the Pivot framework (Volatility, Containment, Reconditioning, Autonomy).
Pathway
The system then identifies the user's stabilization pathway. Pathways determine the context in which guidance is delivered. Examples include physician-guided Suboxone taper, Pivot kratom stabilization protocol, or self-directed detox.
Guidance
Once stabilization state and pathway are identified, the system generates structured behavioral guidance appropriate to that combination.
Client (mobile browser)
↓
Branded check-in form
↓
Data store (Airtable / Sheets)
↓
Automation layer (Make / Zapier / serverless)
Tasks:
compute VDI score + category + trend
evaluate drift rules
generate constrained LLM response based on system state
↓
Guidance displayed to client + operator alert (if triggered)
What I’d Love Your Feedback On
Is the “rule engine + interactive layer” architecture clean?
What would you simplify further?
Are these inputs sufficient for a useful MVP?
What would you build first in 2–3 weeks?
Any obvious pitfalls with this approach?