Enhancing Analyst Efficiency with Gen AI

One of the largest financial providers in the Netherlands sought to explore best practices for developing human-centered Gen AI solutions within their internal ‘Credit Demand Process’. Their objective was to identify opportunities for improved efficiency while ensuring successful adoption by end users (business lenders and product specialists).

We provided our client with the Gen AI Lab offering, 'The Build to Learn Workshop,' designed to foster an experimental mindset for Gen AI innovation. The workshop enabled teams to learn from early failures, pivot quickly, and ensure Gen AI solutions deliver real value. Inspired by Lean UX, it equipped the client team with tools and knowledge to define, develop, and validate Gen AI experiments, including:

• User interviews (conducted beforehand)
• Solution co-creation and prioritization
• Assumption mapping
• Hypotheses building and experiment creation
• Best practices on Responsible AI
• Gen AI Prototyping

The workshop aimed to validate the prototype's potential in streamlining the task of sifting through extensive documents and data to extract relevant information, which is often time-consuming and prone to errors.

Prototype Features

Document Navigation and Summarization:

• During the workshop, analysts used the prototype to scroll through full documents, such as the Cognizant annual report.
• The prototype demonstrated its ability to summarize key strategic and shareholder structures based on references within the document.
• Analysts could click on a reference to move to the corresponding part of the document, enabling verification of the information.

User Feedback Mechanism:

• The prototype included a feature for analysts to rate the usefulness of the references provided by the AI (e.g., "good" or "not helpful").
• This feedback mechanism was designed to help developers understand user preferences and improve the AI's accuracy.

Content Retention Measurement:

• The prototype measured how much of the AI-generated content was retained by the analysts.
• For example, if an analyst removed part of the content, the retention percentage decreased, indicating how much of the AI's output was being utilized.

Workshop Insights - Client Discoveries

• The build-to-learn workshop provided a hands-on opportunity for analysts to test the prototype's capabilities and gather feedback.
  
  
• Analysts asked the prototype to summarize key information from documents and verify the accuracy of the references.
  
  
• The feedback mechanism allowed analysts to provide real-time evaluations of the AI's performance.
  
  
• The content retention measurement feature provided insights into how much of the AI-generated content was used by the analysts.

Prototype Benefits

1. Efficiency: The prototype significantly reduced the time analysts spent scanning through documents.
   
   
2. Accuracy: Automated summarization and reference linking improved the accuracy of the information extracted.
   
   
3. User Engagement: The feedback mechanism ensured that the AI continuously learned and improved based on user input.
   
   
4. Insightful Metrics: The content retention measurement provided valuable data on the prototype's effectiveness and user engagement.

The Results

During the one-day workshop, the client identified 17 opportunities, co-created 4 concept sketches, defined 7 experiments and validation methods, and delivered a stand-alone interactive prototype. The interactive prototype demonstrated the possibility to gather and summarize information while enabling active and passive user feedback on usability and trust.

The workshop is very inspiring and fruitful for our product as well as for our team. The preparation and workshop are tailored for our needs and very well designed. Lots of effort is put in by the experts from Cognizant and the workshop experience is fun, interactive, and innovative. During the workshop, we were able to transform complex challenges into opportunities, identify measurable values, and define concrete experiments leading to actionable next steps. It is amazing that this is doable in a day. After the workshop, we have many inspirational ideas. The team cannot wait to try the experiments out with more users. I highly recommend this workshop. Client’s Team Lead

Interested?

Schedule a tour of our Gen AI Lab or book a workshop for you and your team. Visit our Gen AI Lab page for more information.