The Platform

AI That Works
for Justice

AltReform gives advocates, public defenders, and policymakers the analytical power that was previously only available to prosecutors and prison contractors. Data equality for the people fighting to end mass incarceration.

Reform Score Engine

Our ML model evaluates every reform proposal across four dimensions: recidivism reduction potential, racial equity impact, cost-effectiveness, and political feasibility. Each initiative receives a 0,100 composite score grounded in 20+ years of outcome data.

Machine LearningPredictive ScoringPolicy Analysis
State Reform Tracker

Real-time monitoring of criminal justice legislation across all 50 states and DC. Track active, passed, and proposed reforms. Filter by type, status, score, and racial equity impact. Updated continuously from legislative databases.

All 50 StatesLive TrackingLegislative Data
Racial Equity Analysis

Every reform is scored on its projected impact on racial disparities, incarceration rates, sentencing outcomes, and policing patterns. We model expected outcomes for Black, Hispanic, and Indigenous communities specifically.

Disparity ModelingBJS DataEquity Metrics
Cost-Benefit Modeling

Compare the fiscal cost of proposed reforms against projected savings, reduced incarceration days, lower recidivism, increased employment and tax contribution. Every dollar argument a policymaker needs, backed by data.

Fiscal AnalysisROI ModelingBudget Impact
Advocate Intelligence Briefs

Distilled policy briefs built for public defenders, community advocates, and reform organizations. Download ready-to-use documents with data citations, comparable reforms, and outcome projections for any reform in our database.

Downloadable BriefsCitation-ReadyAdvocacy Tools
Recidivism Risk Modeling

Not individual risk scores, systemic risk analysis. We model how policy changes affect recidivism rates at the population level, helping states design better reentry infrastructure before people leave prison.

Systems ModelingReentry PolicyPopulation-Level

How the Model Works

Our scoring pipeline from raw legislative data to reform intelligence.

01
Data Ingestion

We pull from 40+ public datasets: BJS, Vera Institute, NCSL, state DOC reports, court records, and academic literature. All data is normalized, validated, and version-controlled.

02
Reform Classification

Each initiative is tagged by type, jurisdiction, status, and policy domain. Our NLP pipeline extracts key provisions and maps them to our outcome taxonomy.

03
ML Scoring

Our gradient-boosted model is trained on historical reform outcomes from 1990,2023. Features include legislation text embeddings, demographic context, and prior-state outcomes.

04
Equity Analysis

Each score is disaggregated by race, income, and geography to surface reforms that improve aggregate outcomes while worsening disparities, a pattern invisible to standard metrics.

05
Human Review

Every high-stakes score is reviewed by a criminal justice policy expert before publication. Machine learning informs, it doesn't decide.

Ready to run a prediction?

Start with your state or explore a specific reform type. Our model returns results instantly.

Explore the Map →View Research