Merit Engine - Project Charter

Published

June 29, 2026

internal

Merit Engine: AI Public Safety Promotional Readiness Tutor

Project Charter - Version 1.0

Prepared by: Tonya R. Dawson, Fairlawn Strategy Partners (FSP) Date: June 29, 2026 Status: In Development - Beta Design Phase

1. Executive Overview

The Merit Engine is an IRT-driven (Item Response Theory) AI tutoring platform built specifically for public safety promotional candidates - police officers, firefighters, and command staff preparing for rank advancement exams. It is not a study guide. It is a psychometric intelligence engine that diagnoses an officer’s latent ability (theta, or θ), targets their specific knowledge decay points, and validates their readiness to command before they ever sit for an oral board or written exam.

Core Value Proposition: > “We aren’t helping officers pass a test. We are validating their readiness to lead.”

Business Model: License the platform to public safety agencies (department-wide contracts) at tiered pricing. Where candidates fall short in specific competency areas, offer supplemental e-learning led by Tonya R. Dawson or Caraline Malloy.

2. Problem Statement

What is broken in the current market

Pain Point Current Reality Merit Engine Fix
Content volume Officers study 1,000+ pages of SOPs, penal codes, fire tactics AI prioritizes by criticality weight and knowledge decay
Prep equity Study groups favor “insiders” - an informal inner circle 24/7 AI access levels the field for all shifts and units
Validation gap Standard exams measure test-taking, not leadership aptitude IRT measures latent ability (θ), not raw score percentage
Litigation exposure Unvalidated promotional processes invite EEOC complaints Bias-resistant audit trail is built into every session
Succession planning Chiefs promote who tested well, not who is field-ready Bench Strength Heat Map predicts leadership readiness

3. The Science Foundation: IRT vs. Classical Test Theory

Understanding this distinction is what separates FSP from every competitor.

Classical Test Theory (CTT) - What competitors use

  • Counts correct answers as a percentage
  • Every question is treated as equal
  • Cannot distinguish a “lucky guesser” from a true master
  • Does not adapt to the learner

Item Response Theory (IRT) - The Merit Engine standard

  • Measures the latent trait (θ) behind the answers
  • Every item has a calibrated difficulty parameter (b)
  • The 3-Parameter Logistic (3PL) model filters out lucky guesses via a pseudo-guessing parameter (c)
  • The engine knows not just that an officer got a question wrong, but why and at what difficulty threshold they break down

The practical difference: Two officers both score 80%. Officer A got the hard Use of Force and Search & Seizure questions right and missed easy administrative items. Officer B got the easy items right and guessed through the hard ones. Under CTT: tie game. Under IRT: Officer A is identified as the higher-potential leader. The Merit Engine tells a Chief which officer to promote.

4. Platform Architecture (Conceptual)

Layer 1: Knowledge Ingestion

  • Ingest agency-specific SOPs, General Orders, Penal/Fire Codes, POST standards
  • Tag each policy item with a Criticality Weight (e.g., Use of Force = 10, administrative leave policy = 2)
  • Build a Difficulty Matrix calibrated to the specific promotional exam format

Layer 2: The IRT Engine

  • Adaptive Item Selection: Each question served is chosen based on the candidate’s current estimated θ
  • Real-time θ recalculation after every response
  • False Confidence Detection: Flags items the officer answers quickly but incorrectly (a key I-O insight competitors miss)
  • Spaced Repetition scheduling based on the Ebbinghaus Forgetting Curve - re-tests items at the exact moment cognitive decay is predicted to begin

Layer 3: Scenario Simulation

  • Voice-AI oral board simulation: AI role-plays as a hostile subordinate, angry citizen, or superior officer
  • Decision Latency measurement: tracks the gap between receiving a scenario prompt and committing to a decision (2-second vs. 10-second response is field-significant)
  • Quizzing under artificial stressors (simulated radio chatter, time pressure) to build cognitive load resilience

Layer 4: Analytics Dashboard

  • Candidate View: Personal θ curve, 60-day progress, knowledge heat map, predicted final score range
  • HR/Command View: Force-wide anonymized readiness, unit-level bench strength, policy blind spot reports
  • Predictive Success Score: Final probability of mastery per candidate, per competency domain

Layer 5: E-Learning Bridge (Revenue Lever)

  • When a candidate’s θ falls below threshold in a domain (e.g., Constitutional Law, Budget Management), they are flagged
  • FSP offers targeted e-learning modules taught live by Tonya R. Dawson or Caraline Malloy
  • This converts a software license into an ongoing services revenue stream

5. The 60-Day Command Sprint

Phase 1: Diagnostic Foundation (Days 1-15)

Goal: Establish the ability baseline and identify knowledge decay

Period Focus AI Activity I-O Metric
Day 1 Full Diagnostic 100-question adaptive exam across all SOPs and Law Initial θ Baseline
Days 2-7 Gap Mapping AI identifies “False Confidence” - areas answered fast but incorrectly Discrimination Index
Days 8-15 Foundation Building Personalized high-yield summaries of bottom 20% scoring areas Knowledge Decay Map

Phase 2: Targeted Mastery (Days 16-45)

Goal: Close gaps, build long-term retention

Period Focus AI Activity I-O Metric
Weeks 3-4 High-Yield Drills 15-minute daily micro-quizzes on highest-criticality statutes Spaced Repetition curve
Weeks 5-6 Scenario Simulation Voice-AI role-plays disgruntled officer in disciplinary scenario Behavioral Consistency
Week 6 Stress Testing Quizzing with environmental distractors (simulated radio traffic/noise) Cognitive Load Resilience

Phase 3: Peak Performance (Days 46-60)

Goal: Strategic test-taking stamina and anxiety regulation

Period Focus AI Activity I-O Metric
Week 7 Mock Simulation 4-hour proctored exams with dept-weighted scoring identical to actual exam Predictive Validity
Week 8 Sniper Drills Quizzing only on the final 5% of missed items Mastery Ceiling
Day 60 Taper and Reset High-level philosophy review, go/no-go readiness visualization Peak Readiness State

6. Business Model: Tiered Pricing

Tier Service Level Target Client Price Range E-Learning Upsell
Bronze: The SaaS Access to agency-tuned AI study tutor with basic SOP training Individual officers or small-volume departments $199-$350/user (one-time license) Optional add-on
Silver: Path to Mastery Personalized 60-day plan, weekly AI performance audits, high-fidelity oral board simulations Mid-sized departments (50-200 sworn) $15k-$40k/dept contract Included at trigger thresholds
Gold: Leadership Pipeline Full IRT-validated diagnostic for whole dept, succession planning analytics for Chiefs, “Success Prediction” reports Large municipalities, state agencies $100k+ (multi-year strategic partnership) Included, custom curriculum

The E-Learning Revenue Loop

When a candidate scores below a department-defined threshold in a competency area: 1. Merit Engine flags the gap automatically 2. Candidate is offered a targeted 4-6 hour e-learning module 3. Module is taught live (Tonya or Caraline) or on-demand 4. Completion is documented in the candidate’s Merit Engine profile as a validated remediation

7. Competitive Positioning

Feature All-Write Testing Truleo / PowerDMS NotebookLM Merit Engine (FSP)
Core Function Static exam prep Policy management and analysis Document synthesis chatbot IRT-driven psychometric tutor
Logic Engine Linear quizzing Administrative workflow General LLM Adaptive IRT with 3PL model
Personalization One-size-fits-all Search-based User-fed prompts Theta-calibrated per candidate
I-O Validation Content-based only None (operational) None (general) Criterion validity, bias audit trail
Strategic Goal Passing the test Compliance Information retrieval Predicting and validating leadership readiness
Exclusivity Mass market Enterprise software Public consumer Agency-sovereign instances

Key differentiator: Competitors sell content delivery. FSP sells validated intelligence. Content is cheap. Calibration is expensive. That is the price justification.

8. Go-to-Market Strategy

Primary Buyers

  1. Training Officers / HR Directors - operational pain: inconsistent prep, oral board liability
  2. Police/Fire Chiefs - strategic pain: succession gaps, litigation exposure, retention
  3. Graduate School Program Managers - channel partner: executive leadership programs that train Chiefs

Target Market Entry

  • Pilot: Mid-sized departments (50-200 sworn) at Silver Tier - best balance of revenue and proof-of-concept data
  • SDPD and comparable “Big 10” agencies - aspirational anchor clients that validate the platform nationally
  • University partnerships - embed Merit Engine into executive leadership curriculum as a “Digital Teaching Assistant”

Exclusivity Model (Selective Engagement Policy)

  • Maximum 3 marquee agencies or elite institutions per region per calendar year
  • Agency data is never used to train models for competing jurisdictions (Sovereign Instance model)
  • FSP reserves the right to decline engagements where culture does not align with Ethical AI and Objective Merit

9. Assumptions Being Challenged

The following are working assumptions that should be tested with beta users before hardening the product:

Assumption 1: Departments will share SOP data with a third-party AI vendor

Challenge: Many agencies have strict data governance policies and legal counsel concerns about uploading internal General Orders to an external platform, even with SOC 2 compliance. Mitigation to explore: On-premise deployment option, air-gapped environments, or a “bring your own document” upload model where FSP never retains the raw text.

Assumption 2: Officers will engage with daily 15-minute micro-sessions voluntarily

Challenge: Shift work, overtime, and fatigue are real. Voluntary adoption may be low without departmental mandate or incentive. Mitigation to explore: Gamification layer, shift-aware scheduling that pushes sessions during known downtime, and union engagement strategy.

Assumption 3: IRT calibration is achievable without large historical item-response datasets

Challenge: IRT item calibration requires substantial pilot data (typically 200-500 responses per item) to accurately estimate difficulty parameters. A new agency with limited prior exam data will have uncalibrated items. Mitigation to explore: Use pre-calibrated item banks from POST standards and published public safety exams as a starting scaffold, then recalibrate over time with actual response data.

Assumption 4: Voice-AI oral board simulation will feel authentic enough to build real preparation

Challenge: Current voice-AI has latency, unnatural cadence, and limited emotional range. Officers may dismiss it as a gimmick. Mitigation to explore: Frame it explicitly as a knowledge stress-test (not a human substitute), and measure decision latency and content accuracy rather than conversational naturalness.

Assumption 5: A $15k-$100k price point is accessible within standard department training budgets

Challenge: Many mid-sized departments operate on thin training line items and require city council or grant approval for new technology spend above a certain threshold. Mitigation to explore: Grant writing support (COPS Office, BJA), per-officer pricing at bronze tier as a foot-in-the-door, and framing Gold Tier as litigation risk insurance.

Assumption 6: Tonya and Caraline can personally scale e-learning delivery as a revenue stream

Challenge: Live instruction is the highest-margin service but the least scalable. If 20 agencies trigger the e-learning upsell simultaneously, capacity breaks. Mitigation to explore: Build asynchronous on-demand versions of core modules early. Reserve live instruction for high-complexity domains or Gold Tier clients only.

10. Technical Build Roadmap (Iteration 1)

Phase 0: Foundation (Months 1-2)

Phase 1: MVP (Months 3-5)

Phase 2: Intelligence Layer (Months 6-9)

Phase 3: Scale (Months 10-18)

11. Key Personnel

Role Person Responsibility
Founder / I-O Lead Tonya R. Dawson Industrial Organizational Psychology Professional; psychometric framework, client relationships, e-learning instruction
Co-Founder / E-Learning Lead Caraline Malloy Industrial Organizational Psychology Professional, researcher, and e-learning instructor; leads live and on-demand coaching layer
Software Engineer (TBD) Open IRT engine, adaptive quiz platform, dashboard
Psychometrician Advisor (TBD) Open IRT item calibration, validity studies

12. Success Metrics (Beta Pilot)

Metric Target
Knowledge Acquisition 15%+ average increase in baseline-to-peak assessment scores
Candidate Engagement More than 80% participation rate in daily AI micro-sessions
Administrative ROI Reduction in internal training man-hours by minimum 40 hours per promotional cycle
Predictive Accuracy Predicted score range within +/- 5 points of actual exam score
E-Learning Conversion More than 25% of flagged candidates enroll in supplemental module

13. Open Questions for Beta Users

These are the questions to actively probe during initial client conversations and pilot programs:

  1. What promotional cycle are you running and what is the timeline?
  2. How many pages is your current SOP/General Orders manual?
  3. Do candidates currently receive any department-funded prep support, or is it self-directed?
  4. What is the promotional exam format - written only, oral board only, or combined?
  5. Has your department faced EEOC complaints or grievances related to a promotional process in the last 5 years?
  6. What is your data governance policy for third-party AI tools?
  7. Who owns the promotional process - HR, the Chief, Civil Service Commission, or all three?
  8. What would a measurable “win” look like for you after one cycle?

Document Version 1.0 - June 29, 2026 Fairlawn Strategy Partners, LLC, an affiliate of the Institute for Transformative Change - Confidential and Proprietary Contact: Tonya R. Dawson | tonya@fairlawnstrategy.com