AI Risk Narrative

The AI Risk Narrative is an automated compliance paragraph generated for every wallet screening and comprehensive report. Powered by Claude Sonnet, it produces a human-readable summary of the address's risk profile, counterparty exposure, and recommended actions — ready to paste into SAR filings, compliance memos, or audit documentation.

The AI Risk Narrative is available on all paid plans. It is generated automatically when a screening completes and appears in both the web dashboard and PDF reports.

How It Works

  1. Data Collection — When a screening completes, the platform assembles a structured context package containing the risk score, entity matches, counterparty breakdown, transaction volume, chain activity, and all applicable risk factors.

  2. Prompt Construction — The context is formatted into a structured prompt with strict compliance-writing guidelines. The prompt instructs the model to produce a formal, factual, third-person compliance paragraph.

  3. Generation — Claude Sonnet generates the narrative in real time. The output is a single paragraph (typically 150–300 words) that summarizes the address's risk profile.

  4. Language Selection — The narrative is generated in the language matching the user's locale preference. 13 languages are supported: English, German, French, Italian, Spanish, Portuguese, Dutch, Japanese, Korean, Thai, Polish, Turkish, and Swedish.

  5. Display — The narrative appears in the "AI Risk Narrative" section of the screening result and is included in the downloadable PDF report.

Supported Languages

| Language | Code | Quality | |----------|------|---------| | English | en | Native-quality | | German | de | Native-quality | | French | fr | Native-quality | | Italian | it | Native-quality | | Spanish | es | Native-quality | | Portuguese | pt | Native-quality | | Dutch | nl | High | | Japanese | ja | High | | Korean | ko | High | | Thai | th | High | | Polish | pl | High | | Turkish | tr | High | | Swedish | sv | High |

The language is determined by the user's locale setting in their account preferences. You can change it at any time under Account > Settings > Language.

What the Narrative Includes

Every generated narrative covers:

  • Risk score interpretation — A plain-language explanation of what the numeric risk score means (e.g., "elevated risk due to significant mixer exposure").
  • Entity identification — If the address is attributed to a known entity, the narrative names it and describes its category and threat level.
  • Counterparty exposure summary — Key counterparty categories (sanctioned, mixer, darknet, scam) with approximate percentages.
  • Transaction pattern observations — Notable patterns such as rapid fund movements, high-value transfers, or dormancy periods.
  • Recommended actions — Suggestions aligned with compliance best practices (e.g., "enhanced due diligence is recommended", "consider filing a SAR").

Example Output

The analyzed Ethereum address (0x742d...4a9e) presents an elevated risk profile with a risk score of 72/100. The address has been active since January 2024 and has processed approximately $2.4M in transaction volume across 847 transactions. Analysis reveals that 12.3% of incoming funds originate from addresses associated with known mixing services, primarily Tornado Cash, and 3.1% from addresses flagged as high-risk gambling platforms. No direct sanctions exposure was identified; however, second-hop analysis reveals indirect exposure to OFAC-designated addresses through intermediary wallets. The address is not attributed to a known entity in the BlockchainAnalysis.io database. Given the mixer exposure and indirect sanctions linkage, enhanced due diligence is recommended before establishing or continuing a business relationship with the controller of this address.

Integration with Reports

The AI Risk Narrative is embedded in:

  • Wallet Screening Results — Visible in the dashboard immediately after screening completes.
  • Comprehensive Reports — Included as a dedicated section in the full PDF report.
  • PDF Downloads — The narrative is rendered in the PDF in the user's selected language.
  • API Responses — Available in the risk_narrative field of the screening API response (see API Endpoints).

Accuracy and Limitations

The AI Risk Narrative is a machine-generated summary based on on-chain data and entity attribution. It should be reviewed by a compliance professional before being used in official filings. The narrative does not constitute legal advice.

  • The narrative is deterministic for a given screening result — regenerating it for the same data will produce a consistent (though not identical) output.
  • In rare cases, the model may overemphasize or underemphasize certain risk factors. Always cross-reference with the raw data in the screening dashboard.
  • The narrative does not include off-chain intelligence (e.g., news articles, court filings) unless that information is reflected in the entity database labels.

Next Steps

BlockchainAnalysis.io — Digital Asset Compliance Platform