Eventus launches Frank AI: deterministic AI for financial compliance and surveillance analytics
Eventus announced Frank AI on Oct. 8, 2025 - an AI-driven suite built into its Validus platform for trade surveillance and financial risk monitoring. The focus: deterministic AI that provides secure, repeatable, and transparent results that stand up to regulatory scrutiny.
For product development teams building or buying compliance tech, this is a signal: prioritize determinism, auditability, and data locality over generic text-generation. Frank AI is purpose-built, not a general chatbot bolted onto data.
Why deterministic beats probabilistic in regulated environments
Traditional generative models return probabilistic outputs. That's risky in compliance, where you need consistent, explainable answers. Eventus emphasizes deterministic querying and full audit trails, reducing the risk of hallucinations and easing audits.
Data stays on the secure host, with an on-prem deployment option. That matters for client confidentiality, model governance, and regulator reviews.
What Frank AI includes
- Natural-language querying (chat interface) against live Validus data with deterministic, repeatable outputs
- NLP and LLM-driven tooling integrated into Validus to automate manual tasks (reports, query building, enrichment)
- Behavioral analytics to spot nuanced misconduct patterns across asset classes
- Alert noise reduction and accuracy improvements to streamline investigations
- Full audit trails with structured, fact-based results suitable for regulatory reviews
- Compatibility with leading LLM providers (OpenAI, Anthropic, Google), while maintaining enterprise security controls
- On-prem option and data protection baked into the architecture
Architecture and integration notes for product teams
- Fully integrated with the Eventus Validus platform and trained on Validus-specific data tables for contextual accuracy
- Supports deployment within hours and integrates with existing infrastructure, according to Eventus
- Built to scale for large institutions with upgraded processing capacity
- Data locality: queries and analysis execute without data leaving the host environment
Example workflow
Analyst asks: "Show me all cross-market wash trading patterns involving equity and futures for Client XYZ in the past 30 days, including related party analysis."
Frank AI analyzes multiple asset classes, assembles findings, and returns comprehensive, traceable results. This replaces hours of manual investigation across multiple systems with a single, repeatable request.
What this means for your roadmap
- Shift from text-generation to decision-grade outputs: Prioritize deterministic pipelines that produce consistent, auditable answers.
- Adopt chat-native analytics safely: Natural language can be your interface without sacrificing explainability or control.
- Consolidate toolchains: Embed AI where your surveillance data already lives to reduce integration overhead and handoffs.
- Plan for on-prem and data residency: Keep sensitive data in your environment and document data flows for regulators.
Security and compliance considerations
- Deterministic responses with audit trails and reproducibility
- On-prem option and data never leaving the secure host
- Explainability for model behavior and outputs
- Vendor governance that maps to internal controls and the NIST AI Risk Management Framework
Implementation checklist
- Define target use cases: alert triage, cross-venue pattern detection, entity resolution, reporting
- Map required data tables and lineage in Validus (or your equivalent) to ensure context-rich analysis
- Decide deployment model (on-prem vs. private cloud) and confirm data residency requirements
- Set up access controls, logging, and audit retention policies
- Pilot with a high-friction workflow to measure time saved and noise reduction
- Establish a rollback plan and human-in-the-loop review for critical decisions
KPIs to track post-deployment
- Alert noise reduction (%) and false positive rate
- Time-to-first-finding and total investigation time
- Report generation time and analyst throughput
- Coverage across asset classes and venues
- Number of audit-ready outputs and regulator follow-ups
Why this approach stands out
Eventus applies AI to the actual friction points-querying, cross-asset analysis, and report automation-without giving up repeatability or security. That aligns with how compliance products need to work in production: consistent outputs, tight controls, and clear evidence.
For more information
Explore capabilities and deployment options: Frank AI overview.
Your membership also unlocks:
