How Tencent Leverages RAG and Agents to Supercharge Large Language Models

Introduction

In this article we explore Tencent's large language model applications across various business scenarios, focusing on how cutting‑edge techniques such as Retrieval‑Augmented Generation (RAG) and autonomous agents enhance model intelligence and user experience.

Main Application Scenarios

Tencent's LLMs are used for:

Content generation : ad copy, comment assistance, etc.

Content understanding : text moderation, fraud detection.

Intelligent customer service : knowledge Q&A, user guidance.

Development Copilot : automated code review, test‑case generation.

Role‑playing : NPC interaction in game scenes.

Key Technologies

1. Supervised Fine‑Tuning (SFT)

Fine‑tunes a base model with domain‑specific data, embedding business knowledge to enable targeted task answering.

2. Retrieval‑Augmented Generation (RAG)

Integrates external knowledge bases and retrieval results into the generation process, improving explainability and significantly reducing hallucinations; commonly applied in intelligent客服 and document assistants.

3. Agents

Leverages external tools so the model can perform multi‑step reasoning, planning, and execution, making it suitable for complex tasks that require several inference steps.