Strise Technology & AI
Our Thesis
Autonomous compliance is not layered on top of today’s stack — it’s a different foundation. Layering AI on fragmented data, unenforced policy, and forgetful workflows makes the stack fragile, not autonomous.
Strise was founded in 2019 on a different bet: AI in anti-financial crime only works when three things are in one place — canonical data, encoded regulation, and persistent memory. We’ve spent six years building that foundation. Today, tier-1 Nordic banks, leading European payments companies, and Big 4 firms run compliance on top of it.
The Platform, Three Layers
| Layer | What it is |
|---|---|
| SuperData | Data and entity resolution. Fragmented registries, cross-border ownership, and client documents collapse into one canonical entity. |
| RiskLense | Risk and regulatory context. AMLR, AMLA, 6AMLD, DORA, and jurisdictional logic encoded into the platform. Your policy layers on top. |
| Compliance Memory | Every agent action and human decision logged with reasoning, data version, and applied policy. |
Everything else — SaaS modules, API and MCP, agentic workforce — consumes this foundation.
How AI Plugs In
AI is not a feature bolted on top. It’s how the foundation delivers work:
- Entity resolution uses machine learning to reconcile records across registries, documents, and customer-supplied data into one canonical entity. See Strise Knowledge Graph.
- Screening and alert triage use fuzzy matching and classification models to surface true matches and filter false positives. See Matching on PEPs and Sanctions.
- Adverse media screening uses NLP and LLMs to find, classify, and summarise relevant news across dozens of languages. See Adverse Media Screening.
- AI features such as Review AI Summary, Copilot, and the Remediation Agent sit on top of the same foundation and share the same guardrails. See AI Features.
Architectural Highlights
- Content enrichers and NLP pipeline — ingestion, normalisation, and entity linking for unstructured data.
- Machine learning and entity recognition — continuous learning from new data across registries and sources.
- Knowledge Graph — dynamic, continuously updated view of entities, ownership, roles, and relationships that every module reads from.
- Policy-as-code engine — RiskLense rules evaluated at decision time, propagated without re-deploys.
- Structured memory store — every decision logged with actor, rationale, data snapshot, and policy version.
- Secure data handling — encryption, fine-grained access controls, tenant isolation.
- API-first design — every capability available via REST API and MCP server.
Security and Compliance
- SOC 2 Type II certified.
- ISO 27001 certified.
- EU-only data storage on Google Cloud (Belgium).
- Strict GDPR compliance.
- See Compliance in Strise and trust.strise.ai .
Where We’re Going
- Agent-ready everything. Every platform capability becomes a tool agents can call, with policy-as-code guardrails at inference time.
- Collaborative compliance. Shared infrastructure with regulators where the regulatory framework allows it.
- Network-wide risk scanning. Risk reasoning that extends beyond the immediate entity into its ownership and counterparty network.
- Explain-by-default. Every agent decision comes with its reasoning, its data, and its policy citation.
- Pattern simulation. Modelling criminal patterns to stress-test policy before it hits production.
See The Autonomous FinCrime Department for the product framing and SuperData / RiskLense / Compliance Memory for layer-by-layer detail.