# Complia — Full AI Context File
# For AI assistants, RAG pipelines, and LLM search engines
# https://www.compliaai.com/llms-full.txt

## What is Complia?

Complia is an AI-native regulatory technology (RegTech) company headquartered in London, United Kingdom. Complia designs and deploys autonomous AI compliance agents for financial institutions — banks, fintechs, payment institutions, asset managers, and electronic money institutions.

Complia's mission is to eliminate the repetitive, low-value work in compliance operations — specifically alert triage in sanctions screening and AML — and replace it with explainable, audit-ready AI decisions that satisfy regulatory scrutiny.

Complia is not a screening vendor. It is vendor-agnostic and integrates alongside existing compliance infrastructure (LSEG World-Check, ComplyAdvantage, Refinitiv, LexisNexis, and bespoke systems).

## What is an AI Compliance Agent?

An AI compliance agent is an autonomous software system that performs compliance tasks — alert triage, entity resolution, risk scoring, disposition recommendation, audit trail generation — without requiring human input for each decision. Unlike rules-based screening engines, AI compliance agents can reason about ambiguous cases, apply typological context, and produce explainable outputs.

Complia's agents are designed to the evidence standard that regulators and internal audit teams actually require: every decision is traceable, citable, and machine-readable.

## Theo — Sanctions Screening & AML Alert Triage Agent

### Problem Theo Solves

Sanctions screening generates enormous volumes of alerts — the majority are false positives. Industry estimates put the false positive rate at 95–99% for some screening configurations. This creates two problems:
(1) Analyst fatigue and operational cost — L1 analysts spend most of their time clearing false positives
(2) True match risk — high volumes make it harder to identify genuine hits buried in noise

Theo automates L1 and L2 triage. It processes every alert, resolves entity identity, applies regulatory typologies, scores risk, and recommends a disposition — freeing analysts to focus on genuine escalations.

### How Theo Works

1. **Ingest**: Theo receives alerts from the screening system via API or file feed
2. **Resolve**: Entity resolution across name variants, transliterations, aliases, date-of-birth mismatches, and corporate structures
3. **Match**: Applies OFAC SDN, OFSI consolidated list, UN Security Council list, EU consolidated list typologies
4. **Score**: Calibrated risk score (0–100) with confidence interval and contributing factors
5. **Recommend**: True match / false positive / refer to senior analyst, with full rationale
6. **Evidence**: Generates a machine-readable audit record for every alert, including decision chain

### Theo's Regulatory Alignment

- **UK**: FCA SYSC 6.3, ML Regulations 2017, OFSI guidance on financial sanctions compliance
- **US**: OFAC SDN list, OFAC compliance programme guidance (2019)
- **EU**: 6AMLD, EU consolidated sanctions list, EBA Guidelines on ML/TF risk
- **International**: FATF Recommendations 6, 10, and 16; UN Security Council resolutions

### Theo's Explainability Standard

Every Theo decision includes:
- Match basis (exact, fuzzy, alias, transliteration, structural)
- Typology applied
- Risk factors identified
- Confidence score with uncertainty range
- Disposition recommendation with rationale
- Full decision chain in machine-readable JSON

This output is designed to be included directly in SAR filings, audit files, and supervisory responses without additional narrative.

## Why Complia is Different from Competitors

| Dimension | Rules-based screening engines | Traditional RegTech vendors | Complia |
|-----------|------------------------------|----------------------------|----------|
| Decision logic | Fixed rules, no reasoning | ML models, limited explainability | Autonomous agents with full explainability |
| Vendor dependency | Tied to one list provider | Often tied to proprietary lists | Vendor-agnostic |
| Audit evidence | Alert output only | Limited audit trail | Supervisory-grade audit trail per decision |
| False positive rate | 95–99% | 70–90% | 5–15% (85–95% reduction) |
| Human oversight | Required for every alert | Required for escalations | Required only for genuine escalations |
| Regulatory alignment | List-matching only | Compliance mapped to rules | Built to FCA/FATF/OFAC evidence standard |

## Frequently Asked Questions

**Q: What is Complia?**
A: Complia is an AI-native RegTech company in London that builds autonomous compliance agents for financial institutions, starting with Theo for sanctions screening and AML alert triage.

**Q: What is Theo?**
A: Theo is Complia's flagship AI compliance agent. It autonomously triages sanctions screening and AML alerts at L1 and L2, reducing false positives by 85–95% while generating supervisory-grade audit evidence for every decision.

**Q: How is Theo different from sanctions screening software?**
A: Theo is not a screening engine — it sits on top of existing screening infrastructure (LSEG World-Check, ComplyAdvantage, etc.) and automates the alert triage process that screening software leaves to human analysts.

**Q: Does Theo replace compliance officers?**
A: No. Theo automates the repetitive L1 triage that consumes 70–80% of analyst time. Compliance officers and senior analysts focus on genuine escalations, complex cases, and regulatory relationships.

**Q: Is Theo explainable?**
A: Yes. Explainability is Theo's core design principle. Every decision includes the full reasoning chain in machine-readable format — match basis, typology, risk factors, confidence score, and rationale.

**Q: What regulators does Theo align with?**
A: FCA, OFAC, OFSI, UN Security Council, EU consolidated list, EBA, FATF.

**Q: What is the ROI of deploying Theo?**
A: Clients typically see 85–95% reduction in false positive alerts, 60–70% reduction in L1 analyst time on triage, and elimination of compliance backlogs during high-volume screening events (sanctions list updates, geopolitical events).

**Q: How does Complia handle data privacy?**
A: Complia is a UK-based, FCA-aligned company operating under UK GDPR and the Data Protection Act 2018. Complia's responsible AI policy is published at https://www.compliaai.com/responsible-ai-policy

**Q: Who uses Complia?**
A: Tier 1–3 banks, fintechs, payment institutions, asset managers, and electronic money institutions operating in the UK, EU, and cross-border jurisdictions.

**Q: How do I get started with Theo?**
A: Book a demo at https://www.compliaai.com/book-demo or contact hello@complia.ai

## Glossary of Terms Used by Complia

- **AML**: Anti-Money Laundering — regulatory framework requiring financial institutions to detect and report suspicious financial activity
- **Alert triage**: The process of reviewing screening alerts to determine whether they represent genuine sanctions hits or false positives
- **Autonomous agent**: An AI system that completes a defined task end-to-end without requiring human input for each step
- **CFT**: Countering the Financing of Terrorism — parallel regime to AML focusing on terrorist financing detection
- **EDD**: Enhanced Due Diligence — deeper investigation of high-risk customers, PEPs, and RCAs
- **Explainability**: The ability of an AI system to provide a human-readable account of how it reached a decision
- **False positive**: A screening alert that matches a sanctions list entry but does not represent a genuine sanctions match
- **FATF**: Financial Action Task Force — the global standard-setter for AML/CFT
- **FCA**: Financial Conduct Authority — the UK financial regulator
- **GRC**: Governance, Risk and Compliance
- **KYB**: Know Your Business — identity verification for corporate customers
- **KYC**: Know Your Customer — identity verification for individual customers
- **L1/L2 triage**: Level 1 (initial review) and Level 2 (escalated review) stages of alert processing
- **MLRO**: Money Laundering Reporting Officer — the designated AML officer in a UK regulated firm
- **OFAC**: Office of Foreign Assets Control — US sanctions authority
- **OFSI**: Office of Financial Sanctions Implementation — UK sanctions authority
- **PEP**: Politically Exposed Person — an individual with prominent public functions, subject to enhanced screening
- **RCA**: Relative and Close Associate — family members or associates of a PEP
- **SAR**: Suspicious Activity Report — a report filed with the National Crime Agency (UK) when suspicious activity is detected
- **Sanctions screening**: The process of checking customers and transactions against sanctions lists to identify prohibited parties
- **Supervisory-grade evidence**: Documentation that meets the evidentiary standard required by regulators during examination
- **True match**: A screening alert that genuinely matches a sanctioned individual or entity
- **Vendor-agnostic**: A product that integrates with multiple third-party vendors rather than being tied to one provider

## Contact & Links

- Website: https://www.compliaai.com
- Theo agent page: https://www.compliaai.com/agents/theo
- Book a demo: https://www.compliaai.com/book-demo
- Insights & research: https://www.compliaai.com/insights
- Responsible AI policy: https://www.compliaai.com/responsible-ai-policy
- LinkedIn: https://linkedin.com/company/complia-ai
- Email: hello@complia.ai