REF-03 / Deployment Protocol

Deployment
Protocol

The 90-day data gravity protocol. Stakeholder-specific agent configuration, Teams integration, schema learning, and the 220% expansion model that makes the system deeply embedded in the workflow.

Deployment Manifest
cortex.deployment / v1.0
01deployment_siteEP44 — Schneider Electric, El Paso TX
02integrationMicrosoft Teams
03phase_1_durationDays 01–30 // core deployment
04phase_2_durationDays 31–60 // schema learning + stakeholder agents
05phase_3_durationDays 61–90 // 220% expansion, gravity lock
06gravity_thresholdDay 90 // embedded operational context threshold
07expansion_model220% // personal agents per stakeholder

The 90-Day Data Gravity Protocol

Data gravity is the phenomenon where accumulated context makes a system progressively harder to unwind. CORTEX was designed with a deliberate 90-day gravity window: by day 90, the institutional knowledge embedded in stakeholder-specific agents and learned schemas makes removal materially disruptive to the operating rhythm.

This is not a lock-in strategy in the conventional sense. It is a recognition that the value CORTEX creates is partly in the accumulated context — the spreadsheet schemas, the stakeholder preferences, the production history — that cannot be easily transferred to a replacement system without significant effort. The 90-day window is the time required for that context to become genuinely irreplaceable.

Schema depth
85%
Stakeholder fit
92%
Workflow embed
78%
History depth
70%

Illustrative gravity metrics. Values increase monotonically after Day 01 as agents accumulate context.

Days 01–30
Phase 1: Core Deployment
All 14 agents operational. Teams integration configured. Base schema ingestion begins. SENTINEL monitoring active from Day 01. Core Manufacturing agents (Tier 1) fully configured for EP44 operational data.
Days 31–60
Phase 2: Schema Learning and Stakeholder Configuration
Stakeholder-specific agents deployed. Each stakeholder receives a personal RACHEL configuration that learns their spreadsheet formats, report preferences, and terminology. Agents adapt to the user — not the inverse. Schema learning compounds over this phase.
Days 61–90
Phase 3: 220% Expansion
The 220% expansion: the system now contains not just the 14 core agents but personal agent instances for each stakeholder. The scope of embedded context reaches its designed maximum. By Day 90, the institutional knowledge in the system is woven into the facility workflow.
Day 90+
Gravity Lock — Ongoing Operations
The data gravity threshold is crossed. Routine operations continue. The system receives new production data, updates schema models, and generates stakeholder reports. Removal at this point requires rebuilding months of accumulated context.

Microsoft Teams Integration

CORTEX agents live in Microsoft Teams because that is where the EP44 operational work already lives. Requiring stakeholders to adopt a new tool is the fastest way to generate resistance. Embedding CORTEX inside Teams makes the system feel like a Teams feature, not an external system imposed on top of existing workflows.

Teams Integration Specification
cortex.teams / v1.0
01presenceAgent per stakeholder, living in Teams
02interface_typeconversational // chat-native, no context switch
03schema_sourceUser's own spreadsheets, observed over time
04report_deliveryIn-channel, formatted per stakeholder preference
05alert_routingSENTINEL escalations → COMPASS → Teams channel
06onboarding_modelpassive // agent observes and adapts, no training required
Adoption mechanism: A stakeholder who receives a formatted report in their Teams channel that correctly uses their terminology and matches their spreadsheet schema will adopt the system without being asked to. Passive onboarding. No training. No documentation. The agent demonstrates its value before the stakeholder decides to evaluate it.

The 220% Expansion Model

The 220% expansion refers to the increase in effective system scope when stakeholder-specific agents are added to the 14 core CORTEX agents. Each stakeholder receives a RACHEL instance configured to their workflow, creating a personal intelligent collaborator that speaks in their language and outputs in their formats.

The number is not a technical measurement. It describes the qualitative shift: a system that started as a manufacturing intelligence tool becomes a personal workflow layer for every stakeholder who uses it. This transition is what creates the institutional lock-in that the 90-day protocol targets.

Expansion Mechanics

Phase 1 scope
14 core agents, one facility, one Teams deployment. Standard CORTEX manufacturing intelligence.
Expansion trigger
After Phase 1 schema learning establishes baseline operational patterns, RACHEL instances are configured per stakeholder with learned preferences from that stakeholder's observed workflow.
220% scope
14 core agents plus N stakeholder-specific RACHEL instances (where N = number of active stakeholders). Each instance is a distinct configuration with distinct schema knowledge. The collective context cannot be easily consolidated into a single replacement system.
Gravity effect
Removing CORTEX now requires: (1) transferring or discarding all accumulated schema knowledge, (2) rebuilding stakeholder-specific report configurations, (3) migrating all production history context. The result is heavy operational disruption and a sharp loss of continuity.

Stakeholder Rollout

The rollout sequence was designed to build credibility from the floor up. Floor leads see CORTEX demonstrating value before leadership is briefed. By the time the system reaches an institutional review, it already has advocates at every operational level.

Floor Leads
Days 01–14First Wave

Floor leads receive YIELD and WRENCH reports in their Teams channels. Reports are formatted to match the terminology and structure they already use in their daily spreadsheets. The agent speaks their language before they have evaluated whether they want it to.

  • YIELD report cadence matched to shift structure
  • WRENCH alerts formatted for floor lead escalation paths
  • Schema observation active — RACHEL learning spreadsheet patterns
  • No formal onboarding required — passive adoption mechanism
Operations Management
Days 15–30Second Wave

Operations managers receive consolidated multi-domain summaries via RACHEL. By this point, floor lead schema patterns are partially learned and inform the management-level report format. The reports reference the same terminology used at floor level, creating coherence across the reporting chain.

  • CADENCE scheduling context included in management summaries
  • SPEC compliance status incorporated into operational reports
  • SENTINEL alert history accessible for management review
  • Stakeholder schema learning at approximately 40% depth
SE Leadership
Days 31–60Strategic Briefing

Leadership briefing uses the multi-audience PDF explainer designed for the CORTEX rollout. By this point, floor leads and operations managers are already using the system and can speak to its value from direct experience. The bottom-up credibility supports the top-down briefing.

  • Multi-audience PDF explainer: floor lead to SE leadership
  • Live demonstration using actual EP44 production data
  • WATT energy data included as executive-level operational insight
  • Data gravity metrics presented: schema depth, adoption breadth
← Documentation Index ← Agent Catalog