How a Leading U.S. Precious Metals Investment Firm Saved Over $200K Annually with AI-Powered Compliance Automation
Industry
Finance
Overview
A leading U.S.-based precious metals investment firm helps customers diversify and protect their wealth through gold and silver investments. As part of its onboarding and transaction process, the company is required to record and verify thousands of customer voice confirmations every month to meet strict regulatory and compliance standards. With rising call volumes and increasing compliance complexity, manual verification workflows were becoming hard to scale, slowing onboarding, inflating operational costs, and increasing audit risk.
Challenge
To modernize its compliance operations, the client needed to automate its customer voice verification process without compromising regulatory accuracy. Manual workflows were creating friction at every level. Inconsistent transcription accuracy meant the team was constantly dealing with rework and verification gaps. Compliance data was scattered across systems, making audits a slow and painful process. And without any centralized monitoring in place, issues would often go unnoticed until they had already escalated into bigger problems. Specifically, the client needed to address:
Key objectives included:
- Reducing the time and effort spent manually reviewing customer call confirmations.
- Improving transcription accuracy across different accents, call qualities, and recording conditions.
- Eliminating compliance gaps caused by incomplete or missed confirmation responses.
- Creating structured, auditable records that could hold up during regulatory reviews.
- Scaling with increasing call volumes without growing the operations team in parallel.
- Gaining real-time visibility into compliance status, alerts, and verification queues.
The Solution
Sigma Solve designed and implemented an AI-powered compliance automation platform, tightly integrated with Salesforce, to automate the end-to-end verification process. The cross-functional team, comprising AI engineers, Salesforce integration specialists, QA experts, and project managers, delivered the project using an Agile methodology with iterative sprint cycles and continuous stakeholder feedback. Key components of the solution included:
Core capabilities:
- Whisper AI deployment within a secure Azure Virtual Machine environment, delivering highly accurate speech-to-text transcription even in challenging audio conditions.
- AI-powered entity extraction and compliance validation logic that automatically identified mandatory customer confirmations and verified required script responses, including a minimum number of confirmed "YES" responses before a recording could be approved.
- Salesforce integration through JSON-based workflows, enabling automated alerts, real-time compliance dashboards, and structured reporting.
- Severity-based issue classification to support proactive compliance management, ensuring teams can act on problems before they become violations.
- Django and PostgreSQL-powered backend providing reliable processing, structured data storage, and the architectural flexibility to scale.
Outcome
The results were substantial. Compliance verification turnaround improved by 65%, driven by end-to-end workflow automation that removed manual bottlenecks from the process. Transcription accuracy reached 93% or higher across diverse accents and audio inputs, significantly reducing rework and verification gaps. Faster, more reliable onboarding translated directly into customer experience improvements, with overall satisfaction scores rising by 78%. On the operational side, the platform saves more than 650 manual hours every month, contributing to annual cost savings exceeding $200,000. Compliance teams now have real-time monitoring with proactive severity-based alerts, and every verification is captured in fully auditable, structured records. The cloud-based architecture is designed to accommodate future call volume growth without requiring additional operational headcount.