AI-Based Investment Signal Development and TradingAgent Workflow Orchestration — In Progress

Building on a prior-year proof of concept, the client wanted to move a quantitative investment signal — derived from LLMs and topic models applied to external data — from prototype into live production use across its investment universe. At the same time, the client wanted to explore how specialized, orchestrated AI agents could support its trading workflow.

Templefield Technologies was engaged to bring the prior signal research to production readiness and to design a "TradingAgent" workflow: a set of specialized, orchestrated agents connected to relevant data sources and tested against the existing investment process, alongside continued development of ML/DL models for macro, multi-asset, and selection signals.

• Integrating external data sources into the Bloomberg BQuant environment to support signal generation
• Developing a quantitative investment signal using multiple LLMs and topic models, building on the prior-year proof of concept, and taking it live for the investment universe
• Conceiving and building a "TradingAgent" workflow by setting up, orchestrating, and linking specialized agents
• Connecting relevant data sources and running a test integration into the existing investment process
• Developing and validating machine learning and deep learning models to generate investment signals, with a focus on macro, multi-asset, and selection strategies

This engagement is in progress, running July–December 2026. It builds directly on Templefield's prior NLP-driven signal work for the client, aiming to deliver a live, production-grade investment signal for the client's investment universe and an operational, multi-agent "TradingAgent" workflow integrated into the existing investment process.