CONFIDENTIAL · FOR AUTHORIZED PERSONNEL ONLY · CAIBots © 2022–2026
Executive Demo Brief · KYC / AML Agentic AI
KYC / AML
Agentic AI
Seven specialized compliance agents run CDD, EDD, SAR narrative, and OFAC screening in parallel under a versioned policy gate (KYC-2026.07). BSA Officer Workbench with sealed recommendations, mandatory rationale, dual sign-off on SAR and sanctions decisions, and a three-signature de-risking committee. Demo pipelines run in seconds; production SAR narratives generate in 1–3 minutes inside a governed case.
90-Day Pilot ReadyBSA / FinCEN CompliantSAR-Ready Output7 Parallel Agents11 Scenarios + Live AI
7
Compliance Agents
<12s
Watched Demo Runtime · prod: min–hrs by lane
9
KYC/AML Scenarios (+ Live AI)
<$0.10
AI Cost / Case · design target
Section 01
Access
Demo Credentials
Demo Access
Username: cai or caibots
Password: Demo (capital D)
Live AI: paste sk-ant-… API key
Static mode requires no API key — all 9 scripted scenarios run with pre-built SAR narratives and EDD memos. For live Claude-generated SAR narratives, enter a Claude API key via the ⚙ gear icon in the AI Engine bar at the top of the demo.
Client Data Feeds (optional): Open the Client Data Feeds panel in the left column to upload your own customer history CSV and case archive TXT — the pipeline will use your data. Click SEC EDGAR Live to connect real-time SEC filings, and Actimize / Verafin to load demo AML alert data. Use ⚡ Upload All & Activate to connect all four feeds in one click.
Live AI Mode (Scenario 05): Enter your Anthropic API key via the ⚙ gear icon → Scenario 05 becomes a live investigation. Type any entity name and Claude generates a genuine FinCEN-aligned SAR narrative citing actual FATF typologies and regulatory citations. Cost: $0.14–$0.90 per case (33k–220k tokens at $4.10/M) per investigation.
Section 02
Product Overview
What the Pipeline Does
7 agents run simultaneously: identity verification, sanctions/PEP screening, adverse media, transaction pattern analysis, typology matching, and SAR narrative generation. BSA Officer gets a fully pre-packaged workbench for every HITL case.
ID
Identity & Document Agent
Biometric liveness · Document OCR · SSN validation
SC
Sanctions & PEP Agent
OFAC SDN · PEP registry · Secondary sanctions · FATF
AM
Adverse Media Agent
Dow Jones · OCCRP · Reuters · 50+ media sources
TM
Transaction Monitor
Actimize SAM · Typology library · 30-day lookback
KG
Knowledge Graph Agent
Beneficial ownership · UBO traversal · Entity network
SN
SAR Narrative Agent
FinCEN Form TD F 90-22.47 · Auto-population · BSA style
OR
Orchestrator
Coordinates all agents · Audit trail · BSA/AML logging
Section 03
Scenario Library
11 KYC / AML Scenarios + Live Custom AI
From standard CDD to OFAC sanctions block and live AI investigation. HITL scenarios route to the BSA Officer Workbench. BSA Officer must authorize any SAR filing.
| Scenario | Type | What the Agent Does | Analyst Time | Manual Cost | AI Tokens |
|---|---|---|---|---|---|
SDD Auto-Clear |
AUTO | Consumer onboarding · Low risk · Sanctions clear → auto-approve · Zero analyst touch | 30 min | ~$43 | ~33k tok |
EDD + PEP Hit |
HITL | PEP classification · Source of wealth · Adverse media deep scan → BSA officer EDD approval | 6.5hr | ~$553 | ~220k tok |
SAR · Structuring |
SAR | Transaction structuring · FATF T-04 typology match · SAR narrative generated → BSA officer sign-off | 4hr | ~$340 | ~220k tok |
pKYC · Score Drift |
HITL | Risk profile refresh · Score drift +18pts · New UBO detected → CDD-Rule BOI certification discrepancy · tier-change approval | 3.8hr | ~$323 | ~88k tok |
Correspondent Banking |
HITL | Nested SWIFT chain · FATF grey jurisdiction · BVI UBO · FFIEC Chapter 13 review → EDD escalation + de-risk analysis | 5hr | ~$425 | ~220k tok |
314(a) · BOI (2025 scope) · Circular UBO |
SAR | FinCEN 314(a) match · BOI non-compliance · Circular ownership · TBML → SAR mandatory + 314(a) response + CTA referral | 8hr | ~$680 | ~220k tok |
OFAC SDN Match |
BLOCK | Exact SDN match · Blocking obligation · 10-day OFAC report clock → immediate account block + compliance officer sign-off | 6.5hr | ~$553 | ~220k tok |
314(b) Info-Sharing Request |
SHARE | Inbound 314(b) · Registration gate FIRST (halts if unregistered) · ML/TF scope guard · Safe harbor · SAR-status disclosure structurally excluded → BSA-designee release gate | 2.5hr | ~$213 | ~88k tok |
De-Risking Committee |
3-SIGN | pKYC sustained trigger · License wind-down + 12-mo adverse media · Individualized fair-access basis · Confidentiality-safe exit letter → committee of three, each with rationale | 5.0hr | ~$425 | ~220k tok |
TM Alert — False-Positive Disposition | 1-SIGN | Verafin velocity alert · Structuring hypothesis TESTED and REJECTED (NACHA descriptors, contract reconciliation, 31 CFR 1010.311 non-currency basis, payroll-calendar timing) · Close-with-evidence memo on 5 pillars → KYC Analyst concurrence · Rule tuning queued under SR 11-7, never auto-applied | 2.1hr | ~$179 | ~88k tok |
Sanctions Near-Match — Documented Release | 2-SIGN | 87% name-only fuzzy hit vs SDN · Strong-identifier matrix decides (DOB mismatch 33 yrs, POB/nationality mismatch) → NOT the listed person · Hold enforced until Sanctions Officer + BSA 4-eyes authorization ledgered · List-delta re-screen trigger set | 1.6hr | ~$136 | ~88k tok |
Live Custom AI LIVE |
API | Type any entity name · Claude API generates live SAR narrative with real KYC/AML analysis · Requires Anthropic API key (sk-ant-…) | — | — | $0.14–$0.90 |
Section 04
Demo Walkthrough
25-Minute Demo Flow
Sequence designed for a CRO, CCO, or BSA Officer audience. Adjust for compliance-only vs executive groups.
Open & Orient
Log in. Point to the 9-node pipeline bar. Show the live compliance clock. Point to the Session Stats bar — Cases Processed, HITL Reviews, and Compliance Gate. Introduce the ROI bar showing pipeline runtime and cost-per-case.
"Every customer onboarding and transaction alert routes through this pipeline automatically.""
Run Standard CDD
Scenario 01 (Maria Santos). Run pipeline. Auto-approve in ~5s. Show identity check, sanctions screen, risk score.
"Clean-file retail customers clear in seconds. Your team never touches them.""
Run EDD / SAR Case
Scenario 02 (EDD PEP) or Scenario 03 (Structuring SAR). Walk through adverse media hits, typology match, SAR narrative generation. Show BSA Officer Workbench.
"The SAR narrative is pre-written, evidence-cited, and FinCEN-formatted. Your BSA officer reviews and authorizes — they don't draft.""
Run OFAC Match
Scenario 08 (OFAC SDN Match — Khalid Al-Rashid). Show the BLOCK verdict card, 10-day OFAC blocking report clock (31 C.F.R. §501.603), and the HITL gate. Then optionally run Scenario 07 (314(a)/CTA) to show the 14-day FinCEN response clock.
"Zero false negatives on sanctions. Every customer, every transaction, every time — same ruleset. The 10-day blocking report is auto-drafted and logged.""
Live AI — Scenario 05
Expand Scenario 05 (Live Custom AI). Type the prospect's own company name or a relevant entity. Click Run Live Investigation. Claude generates a live, entity-specific SAR narrative in real time. Requires API key entered via the ⚙ gear icon in the AI Engine bar. In static mode, shows pre-built output with prompt to add key.
"Type any entity name — your company, a counterparty, anyone you're currently reviewing. This is Claude writing a real FinCEN-aligned SAR narrative in under 30 seconds.""
ROI Section
Open Brief Section 05. Walk through manual vs AI cost. Let them enter their monthly case volume.
"What does your current EDD caseload cost in analyst hours per month?""
Audit Tab
Show the hash-chained session ledger on any completed case (⬇ Ledger exports it — the chain recomputes independently). Show SAR narrative export with its grounding-check verdict.
"Every decision logged, timestamped, exportable. Exam-ready on demand.""
Close
Show Efficiency Benchmarks table from Section 07.
"Which typology generates the most false positives for your team right now?""
Section 05
ROI Justification
What the Agent Replaces — Honestly
The agent automates the analytical work phase of each case. It does not replace regulatory submission, account action execution, or human judgment at HITL gates.
✓ Agent Handles (in pipeline runtime)
- ✓ Identity document verification
- ✓ Sanctions / OFAC / PEP screening
- ✓ Adverse media deep scan (50+ sources)
- ✓ Transaction pattern and typology matching
- ✓ Beneficial ownership graph mapping
- ✓ SAR narrative auto-generation
- ✓ FinCEN form pre-population
✗ Remains Human (not replaced)
- ✗ FinCEN SAR e-filing submission
- ✗ Account freeze / close execution
- ✗ BSA officer sign-off (HITL gate)
- ✗ Law enforcement liaison
- ✗ OFAC license applications
- ✗ Customer exit decisions
Analyst rate ~$85/hr blended · BSA analyst · Analytical work phase only (screening + narrative drafting) · Excludes FinCEN e-filing, account action execution, BSA officer review time
| Scenario | Analyst Time | Manual Cost | AI Tokens |
|---|---|---|---|
| SDD Auto-Clear | 30 min | ~$43 | ~33k tok |
| EDD + PEP Hit | 6.5hr | ~$553 | ~220k tok |
| SAR · Structuring | 4hr | ~$340 | ~220k tok |
| pKYC · Score Drift | 3.8hr | ~$323 | ~88k tok |
| Correspondent Banking | 5hr | ~$425 | ~220k tok |
| 314(a) · BOI (2025 scope) · Circular UBO | 8hr | ~$680 | ~220k tok |
| OFAC SDN Match | 6.5hr | ~$553 | ~220k tok |
| 314(b) Info-Sharing Request | 2.5hr | ~$213 | ~88k tok |
| De-Risking Committee | 5.0hr | ~$425 | ~220k tok |
Your Numbers — Interactive Calculator
Annual Hours Freed
—
Annual Cost Saving
—
FTE Equivalent
—
Methodology: Analyst time covers identity screening, adverse media review, typology matching, and SAR narrative drafting only. Excludes: FinCEN e-filing, account action execution, BSA officer review time, law enforcement liaison, and OFAC license applications. Manual cost = analyst time × $85/hr BSA analyst rate. Not a performance guarantee.
Section 06
Audience Customization
Tailoring by Stakeholder
Lead with what each stakeholder cares about. The same demo — different opening 60 seconds.
CCO / Chief Compliance Officer
Lead with SAR obligation accuracy and the OFAC true-match interdiction behavior (never claim zero false negatives — match analysis and human release authority are the control). Show Scenario 08 (OFAC SDN Match) — blocking obligation auto-determined, 10-day reporting clock initiated per 31 C.F.R. §501.603. Emphasize BSA officer HITL gate.
BSA Officer
Show the Officer Workbench directly. Pre-packaged evidence packet, SAR narrative, decision log. They review and authorize — no data gathering. Show the audit tab.
CRO / Risk
Lead with pKYC event-driven monitoring. Show Scenario 04 (pKYC · Score Drift — Meridian Capital) — only score-drift cases reach an analyst. 60–70% workload reduction on periodic reviews.
CTO / Systems
Show the 9-node pipeline and source tool integrations: OFAC API, Dow Jones, Actimize SAM, SWIFT, D&B ORBIS. Discuss the BSA/AML corpus in vector RAG memory.
Section 07
Efficiency Benchmarks
Numbers That Move a Decision
Use these in the close. Conservative figures — defensible in a CFO or COO meeting.
| Metric | Before | After | Reduction | What changes |
|---|---|---|---|---|
| EDD Review Time | 4.5 hours (full manual) | ~8s agent + 55 min officer review | 80% | Pre-built evidence packet — analyst evaluates, not researches |
| CDD Auto-Clear Rate | 0% (all manual) | ≥70% retail auto-clear | 70%+ | Score <30 + sanctions clear + no adverse = zero analyst time |
| SAR Narrative Drafting | 3–5 hours (experienced BSA analyst) | ~10s agent generation (watched demo · production: 1–3 min governed) | 99%+ | FinCEN-formatted, evidence-cited, ready for BSA officer review |
| OFAC False Negatives | Variable (fuzzy match quality) | Zero tolerance · Exact + fuzzy + 2-hop PEP network | 100% | Same ruleset, every customer, every transaction |
| pKYC Workload | Full population reviewed | Event-driven — only score drift ≥15pts | 60–70% | System monitors continuously, humans act only when risk changes |
Section 08
Troubleshooting
Common Issues & Fixes
Login fails
Username:
cai · Password: Demo (capital D). Check for trailing spaces. Press Enter or click the button — both work.Pipeline stalls mid-run
Refresh and retry. All 8 static scenarios are self-contained. If using Live AI mode, verify API key starts with
sk-ant-.BSA Workbench not populating
Wait for the pipeline to fully complete. The HITL workbench populates only after all 7 agents complete — status badge shows COMPLETE and the Evidence tab activates automatically. The HITL tab flashes ⚠ for any scenario requiring BSA Officer action.
SAR narrative blank in Live mode
The narrative streams character-by-character (~20s). Wait for the streaming cursor to stop before evaluating output quality.
OFAC match not showing
OFAC scenario is Scenario 08 (OFAC SDN Match — Khalid Al-Rashid). Run that specific scenario — each scenario targets a different compliance typology.
Audit tab empty
Audit trail populates after pipeline completion. Run a scenario first, then click the AUDIT tab — every agent action is logged with timestamps.
"Which typology generates the most false positives for your BSA team right now?"
Book an Architecture Call →