LG Deploys AI Agents for Retail Investing with LSEG

LG AI Research, the London Stock Exchange Group, and Kiwoom Securities signed a strategic MOU to build an "explainable AI" investment service for retail clients, integrating LG's EXAONE-Business Intelligence into Kiwoom's mobile trading app HeroS# (also referenced as Youngwoongmun S#). The partners will surface stock-level forecast scores together with plain-language explanations and the evidence driving each prediction. The initiative also plans expansion into AI-powered wealth management services and joint marketing to capture retail market share.

The deployment centers on LG's large-scale model family EXAONE and a packaged system called EXAONE-Business Intelligence that orchestrates four specialized agents to mimic a human investment workflow. These agents are described as:

• •a data-collection "journalist" that ingests real-time market data, corporate disclosures and macro indicators
• •an "economist" that produces forecasts and trend projections
• •an "analyst" that isolates drivers, anomalies and explanatory features
• •a decision-maker that synthesizes signals and issues a final score

The product produces both numeric forecasts and narrative explanations, addressing explainability and auditability requirements. The service will present scores with supporting rationale and key drivers directly in the retail UI. LG and LSEG have already commercialized variants, e.g., an "AI-Powered Equity Forecast Score" for institutional users, which suggests the retail build reuses tested models and market-data integrations.

This partnership joins three complementary assets: LG AI Research for model engineering and explainable-agent orchestration, LSEG for global market data and corporate information, and Kiwoom Securities for distribution to a large Korean retail base. The move reflects two broader industry trends: increasing demand for explainable AI in regulated financial contexts and the practical commercialization of multi-agent, multimodal systems in production workflows. For practitioners, this is an operational example of shifting from black-box signals to evidence-backed recommendations at scale, combining model outputs with provenance and narrative layers. It also signals a competitive path for exchanges and data vendors to embed model-driven products within broker interfaces, rather than leaving prediction tooling to third-party fintech startups.

Key implementation details to monitor are the explanation formats (feature attributions, counterfactuals, scenario analyses), latency and retraining cadence for real-time signals, and how the service handles regulatory disclosure and model risk management. Also watch whether Kiwoom and partners expose any developer or API access for institutional users, and how user engagement and performance metrics influence expansion into wealth management.

Executive framing: "For financial AI agents, explainability and reliability are just as crucial as accuracy," said Lim Woo-hyung, head of LG AI Research. Nicolas Falmagne, LSEG senior executive, framed the alliance as a turning point for value creation across the financial ecosystem. The partnership is a practical step toward production-grade, explainable AI within retail trading, not simply a research demo.