Define once, execute deterministically. Graam is an interactive credit analyst that structures deals, projects cashflows, and produces IC-ready outputs — replacing the spreadsheets and one-off models that fragment every credit desk.
CDI files, vendor lock-in, black-box waterfalls. Every desk rebuilds the same infrastructure.
Spreadsheet default models, ad-hoc loss curves, inconsistent assumptions across deals.
Analysts spend days assembling IC memos from disconnected analyses.
How It Works
Upload a prospectus or loan tape. Graam extracts the structure, models collateral, runs scenarios, and produces outputs — IC reports, legal documents, cashflows, and analytics.
Interactive
Graam is an analyst, not a black box. Override its severity, splice two cohorts at month 24, weight CDR by FICO bucket — Graam writes the Python, runs it on your tape, and feeds the result into the next projection. Build performance curves collaboratively, audit the data behind every move, and re-run the entire chain with one prompt.
IC Reports
End-to-end investment committee packages for Auto ABS, RMBS, and CRT.
Deal Structuring
Extract from prospectus or build from spec with WAL validation.
Collateral Analysis
Build projection curves from rating-agency vintage cohorts and EDGAR shelf data.
Rating Analysis
S&P methodology — pool classification, stress testing, implied ratings.
Cashflow Projection
CDR/CPR/severity scenarios, breakeven solver, waterfall execution.
Historical Performance
EDGAR time series, Freddie Mac SFLLD, S-curve analysis.
Model Training
GAM and logit default models trained on historical loan data.
Legal Documents
Prospectus generation from validated deal models.
Tape Stratification
Pool composition by FICO, LTV, rate, term, and geography.
Document Intelligence
Query prospectuses in natural language with inline citations.
Open Source
Graam's cashflow engine is open source. Collateral projections, waterfall execution, and bond analytics — no vendor lock-in.
View on GitHubIC reports on live deals, structuring walkthroughs, and open-source deep dives.
Read the blogBook a demo