V0.9.0

PaddlePaddle/PaddleSpeechV0.9.0Jun 9, 2025by danielcampagnolitg

AI Summary

A massive update involving 79k lines of code added primarily by AI agents. It introduces a new strong static typed design, database implementations, and enhanced autonomous agent capabilities.

Key Highlights

Huge codebase expansion (79k lines) via AI
New strong static typed design shared between frontend and backend
Added Postgres and Mongo database implementations
Image support for autonomous agents
UI for inspecting autonomous agent iteration details

New Features

Image support for autonomous agent
Initial Postgres and Mongo implementations
New strong static typed design
Replaced Aider with TypeScript search/replace editor
Autonomous agent iteration details saved with UI
Code Tasks workflow and vector search

Full Release Notes

Huge release with 79k lines added, 32k lines removed, mostly from Gemini 2.5 Pro/Flash using the platform to build itself, updating the application design and docs for the AI as building this also becomes a blueprint for using AI to build applications.
Some highlights:
- Added image support to the autonomous agent
- Initial Postgres and Mongo (for Firestore Enterprise Edition) implementations with test suites shared between all db implementations
- New strong static typed design with shared interfaces and API definitions between the frontend and backed providing strong compile time checking, which is very useful when the AI is making changes across the front and backend.
- Replaced Aider with a TypeScript search/replace editor which will enable deeper integrations (review etc) into the coding agent.
- Autonomous agent iteration details are saved, with a UI to inspect, for greater visibility/debugging of the agents
- The beginnings of a Code Tasks a 'vibe coding'/AI assisted coding workflow (work in progress) and vector search for enhanced codebase intelligence for the agents.

Docs update coming soon