Enhancing Large Language Models for Finance: Break Through Tech AI Fellows Tackle Reliability and Accuracy with Anote

Enhancing Large Language Models for Finance: Break Through Tech AI Fellows Tackle Reliability and Accuracy with Anote

Generative AI technologies like ChatGPT and Claude have transformed industries, yet significant challenges remain—particularly in high-stakes fields like finance, where precision is critical. While large language models (LLMs) excel at general tasks, they struggle with complex, domain-specific queries requiring numerical reasoning, synthesis, and data extraction. As a result, financial institutions still rely on manual processes rather than AI-powered solutions.

Break Through Tech AI fellows are taking on this challenge in collaboration with Anote, an exciting company specializing in AI-driven solutions for financial services. Together, they are fine-tuning LLMs to improve their performance on finance-specific tasks—an innovation that could boost efficiency and decision-making across the industry.

Why Improving AI for Finance Matters

The financial sector depends on vast amounts of data to drive investment and business decisions. Whether extracting numerical figures, performing calculations, or synthesizing insights, LLMs must provide accurate, reliable responses. However, off-the-shelf AI models often misinterpret numbers, struggle with context, and fail to produce actionable insights—posing a serious challenge when millions of dollars are on the line.

By refining LLMs to handle quantity extraction, calculations, and synthesis, Break Through Tech AI fellows are bridging the gap between AI’s potential and its practical application in financial services. Their work has real-world impact, unlocking new efficiencies and reducing reliance on manual review.

The Approach: Fine-Tuning AI for High-Stakes Industries

To enhance LLMs for financial applications, Break Through Tech AI fellows are employing a structured, data-driven approach:

• Baseline Testing: Fellows first evaluate existing LLMs, such as OpenAI’s GPT models and Anthropic’s Claude, to assess their strengths and weaknesses in financial queries.
• Model Selection & Customization: Using a mix of open-source and closed-source models, fellows experiment with architectures and tools to optimize performance.
• Fine-Tuning with Industry Data: Anote provides proprietary datasets, enabling fellows to train models on real-world financial documents to improve their ability to extract, compute, and synthesize information accurately.
• Performance Benchmarking: Throughout the process, fellows rigorously evaluate accuracy, reliability, and efficiency, iterating on fine-tuning techniques for maximum industry applicability.

Bridging AI Innovation and Business Impact

The AI Program Challenge Projects are more than an academic exercise – they are an opportunity for companies across all industries to gain access to emerging machine learning experts who bring fresh perspectives, technical skill, and research-driven approach to solving real-world challenges

For Anote, the collaboration advances their mission of developing trustworthy AI solutions ensuring financial professionals can rely on LLMs for high-stakes decision-making.

Industry as Co-Developers of the Next Generation of Tech Talent

Break Through Tech’s AI Studio invites industry leaders to actively shape the future of AI talent. By engaging in challenge projects like this one, companies like Anote provide students with hands-on experience, mentorship, and exposure to real-world technical challenges. In return, they gain early access to a strong early tech talent pipeline. 

As AI continues to evolve, ensuring its accuracy and usability will be essential—especially in industries where even minor errors can have major consequences. By embracing their role as co-developers of talent, industry leaders can help close skill gaps and accelerate innovation. 

Learn more about Break Through Tech’s AI Program and join nearly 300 organizations across industries that collaborate with us.