
Anton Antonov, Head of AI and Quant Research at dxFeed, will take the stage in Munich on September 22 with his session, “From Demo to Production: A Year of Agentic AI in Financial Markets.” Drawing on dxFeed’s hands-on experience, Anton will share what it takes to move agentic AI beyond promising demos and into production in financial markets, where accuracy, real-time data access, reliability, and rigorous quality and cost control are essential.
The session will be presented at Data and Agentic AI Day Munich, an event bringing together technology leaders, AI practitioners, and businesses to explore real-world applications of data and agentic AI. The program will focus on practical customer use cases, agentic AI, and sovereign cloud solutions, with insights from AWS and its technology partners.
Generic LLMs present particular challenges in financial markets: they can miscalculate, hallucinate instruments and lack direct access to real-time market data. Anton will explore how dxFeed has approached these limitations through three production use cases: its internal Model Context Protocol (MCP) gateway, which brings market data into AI assistants; SummerFox, dxFeed’s agentic market-reporting solution built on Amazon Bedrock; and Grenadier, the company’s in-house machine-learning system for detecting order-book anomalies across the US equity universe in real time.
The session will go beyond AI experimentation to address the practical decisions behind production systems: when to use MCP, when an agentic architecture makes sense, when a purpose-built model is the better answer, and how to factor in price and performance considerations before it’s too late. Grenadier is a particularly compelling example of that approach — demonstrating why the right AI solution is not always an LLM, while operating in production at just one-fifth of its original AWS cost target.
As financial markets move rapidly from AI experimentation toward real deployment, these questions are becoming increasingly important for firms building technology where incorrect outputs can have real financial consequences.
Join Anton and the dxFeed team in Munich on September 22 to hear what a year of taking agentic AI from demos to production has taught us.
Data and Agentic AI Day Munich
September 22, 2026
Anni-Albers-Strasse 21, 80807 Munich, Germany
Registration is open — we look forward to seeing you in Munich.
Related news:


