How Top Tech Companies Use AI and Automation to Supercharge QA Efficiency – Highlights from the DeWu Tech Salon

Event Overview

On September 22, DeWu Technology hosted the "Quality Technology & AI Efficiency" salon in Shanghai, attracting more than 220 on‑site attendees and nearly 20,000 online viewers. The two‑hour session comprised four technical presentations that explored how leading internet companies apply AI and automation to enhance quality assurance (QA) processes.

1. ByteDance – Intelligent Test Design with Large Models

Speaker Huang Zhihao introduced a workflow that leverages large language models to automatically generate test cases. The talk covered the background of test design, the AI engineering architecture, evaluation metrics, and the efficiency gains achieved by reducing manual test‑case writing and enabling direct conversion of generated steps into automated scripts.

2. Ele.me – Simulation Automation in Financial Services

Speaker Wang Jian (Chenhao) described how Ele.me’s financial middle‑platform adopted a dual‑track simulation framework to address a surge in release frequency and regression gaps. Key points included business‑scenario measurement using MD5 hashes, a hybrid online/offline double‑track validation pipeline, and strategies for handling data dependencies and environment isolation.

3. DeWu – Precise Testing Platform for Targeted QA

Speaker Wang Hongmei presented DeWu’s Precise Testing Platform, which improves test relevance and execution speed. The platform provides change‑recommendation, risk‑recommendation, and H5‑recommendation capabilities, achieves near‑100% precision, and integrates with CI/CD pipelines to reduce manual effort to zero while boosting code‑coverage metrics.

4. Meituan – Gray‑Scale Interception for Advertising Algorithms

Speaker Zhang Qian explained the challenges of maintaining quality in ad‑ranking models and introduced a gray‑scale interception system that automatically validates model updates before deployment. The solution includes multi‑layer model descriptions, fault‑analysis based on historical failures, and a rollout process that blocks problematic changes while allowing safe updates to proceed.

Key Takeaways

AI‑generated test designs can dramatically cut manual effort and accelerate script automation.

Dual‑track simulation and data‑dependency mapping enable reliable regression testing at scale.

Precise testing platforms that integrate risk and change recommendation can achieve near‑perfect test coverage and eliminate manual QA.

Automated gray‑scale validation safeguards ad‑algorithm releases, reducing production incidents.

The salon demonstrated that combining AI, simulation, and automated quality frameworks yields measurable efficiency improvements across diverse product domains.