TLDR: Bloomberg’s AI engineers have been actively presenting and discussing advancements in AI agent technology and tool-calling methodologies at various industry events in 2025. Their focus includes overcoming challenges in scaling agentic solutions for financial applications, emphasizing reliability, and optimizing costs. This ongoing work aims to enhance the precision, comprehensiveness, and speed of AI-driven financial tools.
Bloomberg’s commitment to cutting-edge artificial intelligence was a prominent theme at several key industry gatherings in 2025, where their AI engineers shared insights into their work on agent tool-calling methodologies and the broader challenges of scaling AI solutions. While a specific presentation at ‘ACL 2025’ detailing an ‘improved agent tool-calling methodology’ was not directly found, Bloomberg’s presence and contributions at events like the AI Engineer Summit 2025 and the American Association of Law Libraries (AALL) Annual Meeting & Conference 2025 underscore their significant advancements in this domain.
At the AI Engineer Summit 2025, Anu Koder, Head of AI Engineering at Bloomberg, shed light on the complexities of deploying agentic solutions within a large financial institution. Bloomberg has been developing products with generative AI and agentic features for over a year, encountering hurdles related to performance, particularly precision, accuracy, and factuality. Koder clarified Bloomberg’s terminology, distinguishing ‘tools’ as simpler cognitive architectures from ‘agents,’ which are more autonomous systems equipped with memory and the capacity to evolve. A core tenet for Bloomberg is ensuring ‘precision, comprehensiveness, and speed’ in their AI applications. To achieve this, their approach involves implementing guardrails that check agentic steps post-execution, rather than solely relying on complex prompts to prevent errors. This strategy highlights a pragmatic focus on reliability over mere capability, a point repeatedly emphasized by speakers at the summit: ‘Capability does not mean reliability.’
Bloomberg’s efforts extend to optimizing the economics of AI agents. As inference costs remain substantial at scale, strategies such as fine-tuning and knowledge distillation are becoming critical. The company’s experience, as discussed at the summit, reflects a broader industry trend towards making AI agents more cost-effective while maintaining high quality. This includes leveraging open-source models and continuously refining them based on production signals.
Separately, at the AALL Conference 2025, Bloomberg Law showcased significant innovations in its proprietary AI tools, including advanced generative AI-enhanced features like ‘Bloomberg Law Answers’ and ‘AI Assistant.’ These tools are designed to streamline the legal research process by enabling users to quickly find answers and engage in dynamic dialogues with contextually accurate responses. This represents a major stride in applying AI within the legal sector, demonstrating Bloomberg’s practical application of AI agent principles to enhance professional workflows.
Also Read:
- Bloomberg Law Unveils Advanced AI Tools and Enhanced Resources at AALL 2025 Conference
- AI Agent Startups Attract Significant Venture Capital in 2025
Overall, Bloomberg’s presentations in 2025 reveal a strategic focus on building robust, reliable, and scalable AI agent systems. Their work addresses critical challenges in the deployment of AI, emphasizing the importance of rigorous quality assurance, continuous improvement, and a clear understanding of the distinction between AI’s theoretical capabilities and its real-world reliability.
