Ai

JeecgBoot 3.9.1 Expands AI Agents, Multimodal Creation, and Chat2BI

JeecgBoot 3.9.1 Expands AI Agents, Multimodal Creation, and Chat2BI

JeecgBoot 3.9.1 is a substantial AI-focused release that broadens the platform from basic model integrations into a more capable agent and workflow stack. This version upgrades the underlying AI framework to LangChain4j 1.9.1, adds reasoning model support, streaming APIs, multi-session conversations, file parsing, and significantly expands multimodal features such as text-to-image, image-to-image, AI posters, and OCR. It also pushes JeecgBoot’s AI applications toward a fuller agent model with memory, variables, plugins, workflows, MCP support, and richer business-facing tooling like Chat2BI and AI content generation.

What Changed

The biggest shift in JeecgBoot 3.9.1 is the platform-level AI enhancement. The release upgrades LangChain4j to 1.9.1 and introduces support for reasoning models, while making deep thinking optional instead of enabled by default. It also adds streaming invocation interfaces, multi-session support, MCP over both HTTP and STDIO, and file parsing, which together make the AI layer more practical for enterprise-style application building.

Model and multimodal coverage also expands meaningfully. Qwen now supports parameter tuning and web search, while the release adds support for image models, text-to-image, image-to-image, Claude, VL models, Baidu Qianfan models, and Tongyi Qianwen. This gives teams more flexibility when building AI experiences that combine language, retrieval, and visual generation.

JeecgBoot’s AI applications have also been reworked into more capable agents. The new AI application portal and prompt management features sit alongside support for memory, variables, plugins, workflows, MCP, drawing, and card-based content. This points to a shift away from simple prompt wrappers and toward reusable agent-driven business applications.

Workflow capabilities receive another major boost. The release adds variable extraction, variable aggregation, n8n loop nodes, scheduled triggers, SQL nodes, and knowledge base write-back nodes. It also allows flow duplication and direct invocation of workflows from applications, improving composability for internal automation and data-driven AI processes.

On the user-facing analytics side, AI chat now supports file upload and content parsing, while Chat2BI can generate charts directly from AI chat. Supported chart formats include bar, line, pie, area, radar, gauge, and mixed chart types. Chat2BI also supports natural-language querying, multiple data sources, and chart generation from known datasets, making BI interactions more conversational.

The AI toolbox extends these capabilities into packaged scenarios such as AI poster creation, resume generation, writing assistance, image generation, OCR, product search, and image recognition. JeecgBoot also includes new scenario examples such as image description, product search response, writing assistance, and image recognition to help teams accelerate adoption.

Beyond AI, the broader platform also adds an API signature verification annotation called @SignatureCheck, department short-name support, colored dictionary display for multi-select dropdowns, improved desktop file preview, default Uniapp mobile push integration in the push interface, and upgrades to Jimu Report and Jimu BI Screen to version 2.3.0.

Why It Matters

This release matters because it shows JeecgBoot moving beyond basic AI feature checkboxes and into a more integrated application platform for enterprise AI. The combination of agent capabilities, workflow orchestration, MCP support, multimodal generation, and conversational BI suggests a stronger foundation for building internal copilots, business assistants, document-aware chat tools, and automated operational workflows.

For enterprise software teams, the addition of multi-session chat, file parsing, SQL and scheduling nodes, and knowledge base write-back is especially important. These are the kinds of capabilities that turn AI from a demo feature into something that can participate in actual business systems, data flows, and decision support processes.

The multimodal additions also broaden JeecgBoot’s appeal for customer-facing and productivity use cases. Features like AI poster generation, AI writing, OCR, and image-based applications can help teams package AI into workflows for marketing, operations, reporting, and employee assistance without stitching together as many external tools.

Overall, JeecgBoot 3.9.1 is notable not just for adding more models, but for making the platform more agentic, more workflow-oriented, and more useful for real-world enterprise AI deployment.

Official Source: https://github.com/jeecgboot/JeecgBoot/releases/tag/v3.9.1