Smarter audits at scale: Inside Markerstudy’s GenAI-powered Q&A chatbots - Impetus

Smarter audits at scale: Inside Markerstudy’s GenAI-powered Q&A chatbots

Discover how Markerstudy and Impetus leveraged Retrieval-Augmented Generation (RAG), LLMs, and secure architecture to transform audit operations across financial services

In financial services, auditing and regulatory compliance are mission-critical, but often manual, time-consuming, and error-prone. As regulatory expectations increase and data volumes explode, traditional audit processes can’t keep up.

Enter GenAI-powered Q&A chatbots.

Markerstudy Group, a leading provider of insurance services to over 8 million customers across the UK and Channel Islands, partnered with Impetus to reimagine its audit operations. Together, they developed a Retrieval-Augmented Generation (RAG)-based chatbot powered by Large Language Models (LLMs) to streamline compliance and auditing.

The outcome?

60% reduction in audit response time, improved compliance traceability, and secure, real-time access to policy insights – delivered through intuitive natural language interactions.

Why financial services need AI-driven auditing

Auditors and compliance teams in financial institutions handle vast volumes of complex policy documents, requiring meticulous examination to ensure adherence to regulatory frameworks. Retrieving precise and contextually relevant information is challenging, often leading to inefficiencies, compliance risks, and operational delays.

This is where a Q&A chatbot powered by LLMs and RAG comes into play, offering a transformative approach to financial auditing and compliance. With these, financial institutions can:

  • Automate regulatory audits and compliance checks
  • Deliver precise answers from vast policy repositories instantly
  • Ensure traceable, secure interactions across audit workflows
  • Free up teams to focus on higher-value analysis and decision-making

A recent McKinsey report shows 60% of financial firms are actively investing in AI to modernize risk and compliance operations. Markerstudy Group is one of the early adopters making it work in production.

Challenge: Complexity, compliance and scalability

Markerstudy’s audit team needed a scalable solution to:

  • Extract precise answers from dense financial policy documents
  • Ensure interaction traceability for regulatory compliance
  • Maintain data privacy and security at every touchpoint
    Traditional search methods were slow, and document review cycles were lengthy—delaying audits and increasing operational risk.

Solution: GenAI-powered Q&A chatbot built on RAG & LLMs

Impetus, in partnership with Markerstudy, developed a secure, intelligent Q&A chatbot leveraging:

  • Databricks for vector search and secure data orchestration
  • Langfuse for interaction tracking and audit traceability
  • LLMs (GPT-4/GPT-4o) for generating context-aware, high-accuracy responses

This chatbot integrates directly into existing audit workflows and enables auditors to ask complex queries in natural language—and receive instant, evidence-backed answers.

Architecture Overview

Step-by-step data flow:

  • Document ingestion: Audit policy docs from SharePoint were extracted and converted into structured Markdown
  • Data preparation: These were parsed into key-value pairs and stored in a raw data table
  • Vectorization: The content was embedded into vector format and indexed in Databricks’ vector database
  • Real-time retrieval: A vector search endpoint enabled fast, relevant data fetching in response to user queries
  • GenAI response generation: GPT-4/GPT-4o processed the retrieved context and produced precise, compliant responses
  • Security: Access was secured via Databricks secret scopes and access tokens
  • Traceability: Langfuse logged all interactions to support compliance and audit tracking

The system also supports incremental updates, automatically ingesting and indexing new policy documents to keep the data fresh and compliant.

Business impact

The implementation of the Q&A chatbot delivered significant business benefits:

  • Automation of audit processes: Reduced manual effort in retrieving policy documents and verifying compliance adherence
  • Faster audit completion: The time required for audit reviews decreased substantially, improving operational efficiency
  • Enhanced data security & compliance: The chatbot ensured secure and traceable document retrieval, aligning with industry regulations
  • Improved decision-making: Instant access to relevant audit information empowered auditors with better insights

Why the Q&A chatbot is a game-changer for financial services

The adoption of an AI-powered Q&A chatbot offers transformative benefits for financial institutions, redefining how auditing and compliance processes are managed. Key advantages include:

  • Enhanced audit accuracy and transparency: AI-powered document understanding ensures precise, regulation-compliant responses
  • Dramatically faster audits: What used to take hours of manual review now happens in seconds
  • Scalability and continuous learning: The system is designed to evolve—new policy documents are automatically ingested, indexed, and made available for future queries, ensuring up-to-date compliance monitoring
  • Enterprise-grade security and compliance: From access control to credential management, the platform is designed with compliance in mind
  • Seamless integration with existing workflows: Built to integrate with tools like Databricks and SharePoint, enabling quick adoption without workflow disruption

Conclusion: From manual reviews to machine intelligence

Markerstudy Group’s transformation proves that AI-powered auditing isn’t just possible—it’s practical, secure, and impactful. With Impetus’ expertise in GenAI, LLMs, and RAG-based architectures, financial institutions can now modernize how they govern risk, retrieve policy insights, and ensure compliance—at scale and with confidence.

Is your audit process future-ready?

Let’s explore how Impetus can help you build intelligent, AI-driven governance solutions that deliver measurable outcomes.

Authors

Rohit Sharma – Analytics Engineer, Data Science

Naresh Biloniya – Analytics Engineer, Data Science

Learn more about how our work can support your enterprise