Semantic Search Engine for Risk and Compliance landscape
California, USA
Banks across the world have paid about $321 billion in fines since the 2007-2008 financial crisis as regulators stepped up scrutiny, according to a note by the Boston Consulting Group.
The pandemic has sharpened business focus, now companies seek long term value creation by taking into account the needs of all their stakeholders, and society at large. And now government policy will also make a coordinated push in the same direction.
Client
Regulatory Compliance
CLIENT
Client is a Fin-tech startup in the heart of Silicon Valley, California, with a wide breadth of experience in the GRC (Governance, risk management, and compliance) domain. They are an emerging data intelligence company, with a mission to reduce the compliance cost and also decrease the risk of any non compliance.
CHALLENGE
Build an AI powered tool to closely resemble the practitioner’s approach to regulatory research, to achieve the needs and address the deficiencies in the existing manual regulatory research process.
THE TOP 5 CHALLENGES:
Lack of operational resilience - The expectation of regulators worldwide to impose more requirements on firms to strengthen the industry’s overall operational resilience is high.
Manually identifying authoritative sources - Manual search within 900+ regulators, and 300+ million pages of rules. Identify specific rules within each authoritative source, is a complex and time consuming task.
Ever changing complexity of rule - Complexity of the rules further increases the time spent to identify relevant rule and manually map then to policies, procedures, and controls.
Inconsistent - Results are dependent on practitioner expertise
Lacks of accountability - Current trend of focusing more on the regulator's end rules, but not on the process adherence.
SOLUTION
Our main task was to design and develop a semantic search platform that enables Risk experts search across regulations, legislation and industry standards to identify similar requirements. AI powered search uses high-dimensional vectors and graph technology to fetch rules that are semantically similar to the given search phrase.
Scan authoritative sources for changes or new rules, fetches rule changes and highlights the rule text that has changed. Similarity scores between high-dimensional vector representation (sentence encoding) of rules, obligation statements, policies and controls, powers AI to determine the exact impact of rule changes across the business locations.
Government compliance source
eCFR | US CODE | FR | PRIVACY | GDPR | CCPA | NIST ..
Document Parser
Rule Extraction
Classifications
product classes | rule type | rule matters | regulatory type
Unified Data Lake for compliance information
Regulatory search
Semantic search - Search across regulations, legislation and industry standards to identify similar requirements.
Semantic search based on sentence embedding vectors and graph technology, to reduce domain noise.
Search refinement - Regulatory filters on applied classification (rule type, matter type and product class) to refine the regulatory search.
Unified Data Lake for compliance information
Result analysis - Interactive reports and archive search for future reference
Chain Management - Scan authoritative sources for changes and new rules. Highlight he rule changes for impact assessment across business locations.
RESULT
The year 2020 laid the foundation for digital technology and its regulators. Digital financial services are clearly impacting banks and they are forcing banks to rethink and restructure their approaches to becoming resilient, adopting new technologies, leveraging data and constructing agile operating models due to security concerns.
In this effort, we have built a semantic search platform for client to enable them to search rules across the ever changing regulatory sources, and effectively map them to the internal policies, procedures, and controls