Commerzbank’s Generative AI Integration
Staying competitive in today’s fast-paced financial market requires innovation and efficiency. Leading German bank Commerzbank saw the potential to improve its internal operations, particularly in its corporate financial advice section. Sales advisors must document investment proposals in precise methods due to regulations. It requires manual labor and time. This has caused production bottlenecks and reduced client advice time.
Commerzbank worked with Google Cloud to simplify this laborious procedure with powerful AI. Commerzbank automated this procedure and increased sales advisor efficiency using Google Cloud’s AI and machine learning. Financial advisors are vital to client-specific financial advice. Reviewing client contacts and extracting and summarizing domain and client-specific information was laborious and wasteful. Sales advisors spent a lot of time listening to call records, identifying critical points, and manually entering data into systems. This method took time and increased errors and inconsistencies.
Commerzbank’s comprehensive financial advice process automation system uses Vertex AI’s multi-step gen-AI architecture for quality, scalability, and extensibility. Unlike other models, Gemini 1.5 Pro’s capacity to interpret several modalities and assimilate extended context information made this system conceivable. Codelab contains a sample system implementation. The main steps are listed below:
Read: Top 10 Trends Of Customer Experience: Why CX is the Cornerstone of FinTech Success?
Step 1: Sales advisors use a user-friendly frontend interface (1) to select client calls for processing. This interface talks to a workflow-managing Java Spring backend (2). The backend imports the selected audio recordings from Commerzbank’s recording system (3) into GCP storage buckets. This makes data available for AI processing pipelines.
Audio chunking and storage (4.1, 4.2): Segmenting imported audio recordings into smaller, manageable pieces (4.1) helps manage lengthy client calls. The system efficiently processes multi-hour calls using “chunking”. Next, GCP storage (4.2) secures these audio chunks for data durability and accessibility.
3. Advanced diarization and transcription (4.3): This phase is essential for producing a structured transcript that accurately captures conversation subtleties. Gemini 1.5 Pro creates a diarized transcript that appropriately credits each speaker. This technique processes audio chunks sequentially. The model receives the transcript up to that moment, carefully crafted cues, and a few-shot audio-to-text transcription example to maximize accuracy. This guarantees the final transcript is accurate in content, speaker identity, and numerical information, which is vital in finance. To optimize storage, step 4.2 audio portions are destroyed when the final transcript is generated.
4. Fact extraction (4.4): Gemini 1.5 Pro analyzes and extracts relevant information from a lengthy transcript. This involves finding crucial information for the financial advisory document to be finished. The model is asked to identify client identities, investing preferences, risk tolerance, and financial goals.
5. Summary generation (4.5): Creates short and accurate summaries for each area in the financial advisory document. Based on the retrieved facts from the preceding stage and Chain-of-Thought (CoT) prompting, Gemini 1.5 Pro generates numerous German-language summaries for each form field and topic. This makes the summaries useful and compliant with Commerzbank’s internal and regulatory rules.
Read: Identity Fraud: How Much Banks Lost in 2024?
6. Summary optimization and explanation (4.6): The Vertex AI Gen AI Evaluation Service selects the optimum summary for each form field, ensuring high-quality output. The service also provides a human-readable explanation for its selection, helping sales advisors trust the AI and comprehend its reasoning.
Commerzbank can automate a difficult and time-consuming procedure with great accuracy and efficiency using Gemini 1.5 Pro and Vertex AI’s multi-stage design. Commerzbank streamlines these operations to let sales advisors focus on higher-value duties, enhancing client service and business growth. AI-powered automation has had a big influence. Commerzbank increased productivity by automating financial advisory documents. Sales advisors now have more time for client connections and tailored financial advice.
Industry Comments
“Therefore, we set up a Strategic Initiative Corporate Clients AI powered sales force – to make our sales focus on high value activities” Sebastian Kauck, CIO Corporate Clients at Commerzbank.
FAQ
1. What challenges did Commerzbank face in its financial advisory workflows?
Commerzbank’s sales advisors spent excessive time manually documenting investment proposals to meet strict regulatory requirements. The process involved listening to lengthy client calls, extracting key information, and summarizing data, leading to inefficiencies, bottlenecks, and reduced time for personalized client interactions.
2. How did Commerzbank address these challenges?
The bank collaborated with Google Cloud to automate the financial advisory process using Vertex AI and Gemini 1.5 Pro. This system streamlined call transcription, fact extraction, and summary generation, significantly improving efficiency while maintaining accuracy and compliance.
3. What role does Gemini 1.5 Pro play in the automation process?
Gemini 1.5 Pro powers key stages of the process, including:
- Advanced diarization and transcription: Ensures accurate speaker identification and content transcription.
- Fact extraction: Identifies critical client details such as financial goals and risk tolerance.
- Summary generation and optimization: Produces concise, regulation-compliant summaries and selects the most suitable versions.
4. How does the system ensure accuracy and compliance?
The multi-step process uses advanced AI models to generate high-quality transcripts and summaries. It incorporates Chain-of-Thought prompting and Gen AI Evaluation to ensure outputs meet regulatory and internal quality standards.
5. What are the tangible benefits of this automation?
- Increased productivity: Automation saves time for sales advisors by reducing manual tasks.
- Enhanced client service: Advisors can now focus on personalized financial advice and building stronger client relationships.
- Regulatory compliance: AI-generated outputs adhere to strict financial documentation requirements.
6. Is the system scalable and extensible for future needs?
Yes, Vertex AI’s architecture and Gemini 1.5 Pro’s capabilities allow the system to scale as data and client needs grow, ensuring long-term adaptability and efficiency.
Conclusion
Commerzbank’s integration of generative AI, particularly Gemini 1.5 Pro and Vertex AI, marks a significant leap in automating financial advisory workflows. By addressing inefficiencies in call processing and documentation, the bank has empowered its sales advisors to prioritize high-value client interactions while ensuring compliance with regulatory standards.
This initiative demonstrates how AI can transform traditional banking operations, enhancing productivity, accuracy, and client satisfaction. As financial institutions continue to embrace AI-powered solutions, the industry is poised for a future of smarter, more efficient workflows that drive both growth and innovation.
Read: Fintech Marketing: Top 10 Power Strategies to Accelerate Growth
Read: Top 10 Strategies for Effective Fintech Branding
Thanks for reading!
To share your insights with the FinTech Newsroom, please write to us at news@intentamplify.com
