The Evolution and Intelligent Roadmap of Alibaba's Mobile Development Platform mPaaS

At the 2019 Alibaba Cloud Summit in Beijing, Ant Financial technology expert Fu Haitao presented “The Intelligent Road to the Next‑Generation Mobile Development Platform mPaaS”.

The talk covered four main topics: the evolution of the mPaaS platform, the product system of mPaaS 3.0, the parallel evolution of Alibaba’s financial business, and the integrated intelligent mobile scenarios enabled by AI.

1. Development history – mPaaS 1.0 (Dec 2016) focused on extending Alipay’s financial‑grade SDKs and UI components to help financial institutions build stable apps quickly. mPaaS 2.0 added a release platform, hot‑fix, offline packages, data sync and custom analytics to support rapid iteration and digital transformation. mPaaS 2.0 also extracted the Mini‑Program framework for broader use.

2. mPaaS 3.0 product system – The platform now consists of three layers: (a) a flexible front‑end capability offering Native, H5 and Alipay Mini‑Program frameworks, 100+ UI controls and 20+ SDKs; (b) a robust mobile‑mid‑platform that manages the full app lifecycle (development, testing, release, analysis, operation); (c) a stable back‑end connection service providing mobile gateway and large‑file channels.

3. Evolution of Alibaba’s financial business – Similar to the platform, financial services moved from stability‑oriented 1.0 to data‑driven 2.0 and finally to intelligent 3.0, where data ingestion, analysis‑decision engines and mPaaS scenarios form an end‑to‑end mobile AI solution. This enables “千人千面” (personalized) services such as intelligent prediction, smart delivery and OCR.

4. Integrated intelligent scenarios – By leveraging Ant Financial’s AI models and the data collected by mPaaS, the platform can dynamically create user groups for smart prediction (e.g., 7‑day activity or churn), combine with gray‑release, push notifications and ABTest for targeted marketing, and provide smart delivery modules (smart素材, smart圈人, smart推荐, smart监控). Lightweight AI models also power client‑side features such as AR red‑packet recognition.