Introducing AICIL: The Compliance Intelligence Layer
Today we are introducing AICIL -- the AI Compliance Intelligence Layer -- a system that generates complete, regulator-ready compliance documentation packages for SWIFT cross-border wire transfers before the payment is sent. AICIL is built by DeepHavn, the product arm of DeepHavn, our research organization focused on tacit knowledge in financial compliance.
The Problem We Are Solving
Cross-border wire transfers are the circulatory system of the global economy, moving trillions of dollars daily through the SWIFT network. But every payment carries a compliance burden. Banks must verify that the originator and beneficiary are not sanctioned, that the source of funds is legitimate, that beneficial ownership is transparent, and that the transaction complies with the regulations of every jurisdiction it touches.
Today, this verification happens reactively. The payment is initiated, enters the correspondent banking chain, and at each stop, compliance teams independently assess the transaction. If documentation is missing, the payment stalls. If a false positive is triggered -- and 80 to 95 percent of compliance alerts are false positives -- analyst hours are consumed resolving it. The result is $40 billion in annual compliance costs, 3 to 5 day settlement delays, and a steady erosion of correspondent banking relationships worldwide.
How AICIL Works
AICIL inverts the compliance workflow. Instead of assembling documentation after a flag, the system generates a complete compliance package before the wire is submitted. The pipeline has four stages.
First, the originating bank submits the transaction details through AICIL's API -- originator, beneficiary, amount, currency, purpose, and supporting documents. Second, AICIL screens the transaction against all active sanctions lists (OFAC, EU, UN, HMT) in real time, resolving entities against structured databases and producing a signed screening certificate. Third, AICIL assembles the compliance package -- combining screening results, beneficial ownership verification, source-of-funds documentation, and risk scoring into a structured, machine-readable document. Fourth, the package is delivered alongside the SWIFT message, available to every correspondent bank in the chain.
The Research Behind It
AICIL is built on the OGEE framework (Observable, Generalizable, Experiential Elements), developed by DeepHavn. OGEE is based on a simple insight: compliance expertise is tacit knowledge. Experienced compliance officers make judgements based on patterns they have internalized over years of practice -- patterns that are difficult to articulate as explicit rules but that consistently produce correct outcomes.
OGEE captures this tacit knowledge by training on outcomes rather than labels. Instead of asking compliance analysts to label documents as adequate or inadequate, we observe what happens when documents enter the correspondent banking network. Packages that are accepted across multiple jurisdictions encode the tacit standards of the network. Packages that are delayed or returned reveal gaps.
What We Are Building Toward
AICIL is currently in demonstration stage. The system processes test transactions through the full pipeline -- screening, assembly, delivery -- and produces compliance packages that meet the documentation standards of major correspondent banks. We are working with early design partners to validate the package format and integrate with existing payment-initiation workflows.
Our goal is a world where compliance quality improves with every transaction processed. Each payment that flows through AICIL enriches the model, refines the per-bank reward profiles, and improves the accuracy of chain prediction. This is the data-moat thesis: the system gets better as it scales, creating a compounding advantage that is difficult to replicate.
Founded in San Francisco
DeepHavn is headquartered in San Francisco, with the DeepHavn research arm focused on tacit-knowledge modeling and outcome-based training methodologies. We are a small team with deep experience in financial infrastructure, machine learning, and regulatory technology. If you are a bank, payment processor, or fintech working on cross-border payments, we would like to hear from you.