← Cases
United KingdomFinance and insurance2026 – ongoing

Elliptic

AI copilot for crypto compliance analysts

Elliptic's copilot cut Suspicious Activity Report time by 40% for crypto compliance analysts

WorkingVendor case studyVendor ↗

Problem

Elliptic's clients, including financial institutions and government agencies, investigate illicit activity across digital asset transactions. Analysts had to read through wallet histories and write risk narratives by hand.

Approach

Elliptic built a copilot with Databricks Agent Bricks. It summarizes wallet activity and transaction history, points out exposure paths and drafts risk narratives for reports to fit each client's risk appetite. The team logs each interaction with MLflow tracing and scores answers for correctness, relevance and safety with LLM judges and custom safety checks.

Results

2xfaster resolution of high-risk screenings, per DatabricksVendor case study[1]
40%less time to complete a Suspicious Activity Report, per DatabricksVendor case study[1]

Takeaway

Logging every interaction and scoring answers automatically gave the team a way to check a copilot used in a regulated setting.

Categories

Business function
Finance, risk and legal
AI technology
Generative AI
Evidence
Vendor case study

Sources

  1. Protect crypto assets with governed AI
    Databricks23 Apr 2026Vendor publication

Source summaries

[1]Databricks23 Apr 2026

Databricks describes a compliance copilot Elliptic built on Databricks Agent Bricks for blockchain analysts. It describes the evaluation and tracing setup and reports twice-as-fast resolution of high-risk screenings and a 40% cut in Suspicious Activity Report time.

Full details are in the original sources.

Added 4 Nov 2026 · Updated 4 Nov 2026
Open case page ↗