Autographer · Explainable Fraud-Control Discovery

Catch more fraud. Raise fewer alarms.

Autographer mines your own history for the fraud patterns your models miss, then hands your team evidence-scored control candidates with estimated capture, alert load, and review context.

weeks of analysis in one run estimated alert load before testing raw source data is not retained by default after the run
Autographer · Evidence Canvas discovery run: enrich → mine → bundle → model → guard
DATASET
transactions.csv
132,000 rows · 12 fields
is_fraud 27% · Transaction Monitoring
DISCOVERY RUN
job_7f2c
supervised · temporal features on
raw source data not retained by default
amount > 50000sup 42 · prec 0.90
channel = AGENTsup 38 · prec 0.86
beneficiary_is_new = truesup 28 · prec 0.78
count_1h_by_entity > 3sup 24 · prec 0.74
geo = high-risk bucketsup 19 · prec 0.72
CANDIDATE · PRIMARY
Agent-channel burst to new beneficiary
confidence 0.84 diversity 0.91 all predicates grounded
High value agent transfer (w 4)
amount > 50000 ∧ channel = AGENT
New beneficiary burst (w 3)
beneficiary_is_new ∧ count_1h > 3
Shadow test Model editor Open case Deploy · human-approved only
CANDIDATE · ALTERNATE
New location, high-value transfer
confidence 0.58 diversity 0.68 1 guard warning
Location and amount (w 4)
geo_location = high-risk bucket
CANONICAL_PREDICATE_MISSING
amount > 50000 (sup 42 · prec 0.90) was mined as strong evidence but dropped from the generated model. Fix before testing.
Autographer · Evidence Canvas, in the live product
Autographer Evidence Canvas in the Loci console showing a candidate control with canonical predicates preserved from mined evidence, confidence and diversity scores, runtime handoff guard, and shadow test governance exits

Not a mockup: the Evidence Canvas as analysts see it, candidate review panel and all.

Weeks of analysis, one run

The pattern hunting your analysts do by hand, across amounts, channels, beneficiaries, velocity, and geography, runs in minutes, with the statistics attached.

Alert fatigue becomes a choice

Every candidate estimates alert load next to expected capture. Your team can test the trade-off before it hits the queue.

New typologies, days not quarters

When losses spike in a new attack vector, run discovery on the fresh data and ship a reviewed control the same week.

Regulator-ready by default

Every control traces to evidence in your own data: support, reliability metrics, lineage, approvals, and labeled-outcome metrics where available.

The Problem

Generic AI can suggest fraud patterns. It can't defend them.

Pattern discovery today sits between two bad options: manual analysis that takes weeks, and generic AI that produces answers no reviewer can verify.

Generic AI

Fast, but unaccountable

  • Suggestions sound plausible, but carry no evidence you can verify.
  • No support, reliability, or outcome context. Nothing useful to show a model-risk reviewer.
  • The output isn't deployable logic; it still has to be rebuilt by hand.
  • Same prompt, different answer every time. Nothing is reproducible.
  • Sensitive transaction data has to leave your environment.
With Autographer

Evidence-graded discovery, in your environment

  • Every candidate traces to mined patterns with support and reliability metrics; labeled data adds precision and recall.
  • Estimated alert load is shown before shadow testing: expected capture and review cost, side by side.
  • Output is review-ready control logic for shadow testing and governed deployment.
  • Deterministic by design: same data, same patterns, every run.
  • Runs where your data lives. Raw source data is not retained by default after the run.
How It Works

Discover. Review. Ship.

Autographer drafts; your team decides. Human-in-the-loop by design.

1

Discover

Upload a file or connect a dataset. With labels, it learns from outcomes; without them, it hunts for suspicious structure. Raw source data is not retained by default.

2

Review the evidence

Open any candidate in the Evidence Canvas: source patterns, support, reliability, labeled-outcome metrics where available, and generation warnings. Approve, adjust, or reject.

3

Ship

Send approved candidates to shadow testing in Loci's decision engine (MADIE) or your existing workflow. Deployment stays governed and logged.

Capture vs. Cost

See the capture and the estimated alert load before testing.

Every candidate is ranked by estimated fraud capture, alert load, and evidence quality, and traces back to the patterns in your data. Production deployment remains governed and reviewable.

Control candidate Fraud capture False-positive rate Evidence
Block cards with prior chargeback history 87% 0% see evidence →
Flag high-risk email domains 63% 2.1% see evidence →
Review orders > $10k from high-risk regions 41% 4.3% see evidence →
Trust, Engineered

It checks its own work, then keeps checking.

Generated logic is proofread against the mined evidence before review, and reconciled against real-world outcomes after deployment.

Generation Guard
Before review

An automatic proofreader compares every generated model against its evidence, and flags dropped clues, dead clauses, and weak support.

▲ CANONICAL_PREDICATE_MISSING
amount > 50000 dropped from model
▲ DEAD_CLAUSE · condition never matches
● all other predicates grounded
Outcome Reconciliation
After deployment

Predicted performance is tracked against observed reality, precision, alert rate, and drift, so a control that decays gets caught, not trusted.

precision · pred 0.84 vs obs 0.78 (Δ −0.06)
alert rate · pred 3.5% vs obs 3.1%
● tracking · mature after 3 windows
▲ drift watch if deltas widen
Pattern Intelligence
Across runs

Patterns that persist across repeated runs become champions. Autographer tracks how their reliability metrics, outcome-backed precision where available, and thresholds evolve over generations.

amount > threshold · gen 4 · improving
precision 0.71 → 0.79 → 0.86 → 0.90
threshold drift 30k → 42k → 50k → 54k
channel = AGENT · gen 3 · stable
Built to Defend

Walk into model-risk review with the evidence ready.

Black-box vendor findings, fair-lending scrutiny, audit requests, everything a reviewer will ask for is generated as a by-product of how Autographer works, not reconstructed afterwards.

Checked on held-out data

Separate search and validation windows show how a candidate behaves out of sample, not just on the data that suggested it.

Estimated alert load up front

Every candidate carries its alert load, not just its fraud capture, so alert fatigue is a decision, never a surprise.

Reproducible by design

Deterministic discovery: the same data produces the same patterns, every run. No agent improvisation in the loop.

Full lineage

Run, source patterns, evidence state, and approvals recorded for every control, defensible end to end.

One Engine, More Than One Fraud Problem

Prove value on one surface. Extend to the next.

Domain profiles adapt the same discovery engine to new fraud surfaces, new features and vocabulary, not new projects.

Transaction monitoring

Money movement, beneficiary, velocity, and location signals across payments and transfers.

IVR & contact-center

Call, verification, reset, and account-compromise patterns from contact-center event streams.

Refund & tax abuse

Repeatable abuse patterns surfaced from profile-specific features across claims and refunds.

Autographer discovers and drafts Loci decision engine or your existing decision engine your engine decides. Always.
Get Started

Bring one dataset. See the controls it finds.

In 30 minutes, watch patterns in your own data become review-ready control candidates, evidence and all. Discovery only; raw source data is not retained by default after the run.