Live demo
17k

Built an AI Multimodal R&D Platform in Days — with NocoBase

NocoBase empowers Wasu Media to efficiently build an AI multimodal platform, enhancing workflow management and improving content generation efficiency.

Deng lijia |

Introduction

Wasu Media has built an AI multimodal R&D platform from scratch — in just a few days — using NocoBase.

As a major player in digital television and media industry, Wasu Media has been actively exploring how emerging technologies like AI and AIGC can reshape content production.

Here’s how their team turned complex data pipelines and model workflows into a unified, visual platform for AIGC innovation.

KlingAI.png

Image generated by AI

Scenario

In reality, content generation isn’t just about chaining models together — it’s about managing tons of data moving through different steps.

To handle this complexity, the team identified three key areas to focus on:

  • Data Management: Internal multimodal assets (such as text, images, and other materials) must be centrally organized, labeled, and incorporated into a structured content repository. Only with a systematic data management framework can semantic search, content recommendation, and intelligent generation be effectively supported.
  • Workflow and Production Scheduling: From topic selection → content generation → review → output, every stage needs to be automated and visualized. A unified scheduling platform not only improves efficiency but also enables multi-model orchestration, task assignment, and version control.
  • Data Monitoring: In large-scale content generation, real-time monitoring of data—such as scripts, generation records, and model call logs—is critical. It helps teams quickly locate issues and provides valuable feedback for future optimization and iteration.

Challenges

As the team worked to validate workflows quickly while ensuring long-term scalability, several key challenges emerged:

  • Too many moving parts: The number of assets exploded, from text scripts to image outputs. Managing them all — along with models and compute nodes — became chaotic.
  • UI bottleneck: Each interface change required days of front-end work.
  • Fast iteration pressure: Business teams needed new features every few days. Any delay slowed experiments and validation.

Before adopting NocoBase, the team had tried other AIGC-oriented application development tools. However, while those tools offered more vertical features, they still required substantial coding and came with steep learning curves — making them ill-suited for a fast-paced, trial-and-error development process.

Adoption and Implementation of NocoBase

To tackle these challenges, the R&D team ultimately chose NocoBase as the foundation of their system. Its no-code architecture allowed developers to focus on data modeling and business logic, without spending time on front-end development. For a back-end–oriented team, this meant they could get started quickly and dramatically shorten their development cycle.

With NocoBase, the team built their core platform in just 1–2 days — what used to take dozens of APIs and multiple front-end pages.

Workflows replaced hundreds of lines of logic with clear visual automation.

Back-end developers no longer had to touch front-end code — NocoBase automatically generated the interfaces they needed.

Complex business logic was implemented through workflow configurations, effectively turning dozens of lines of if-else statements into clear, visual flowcharts—with alerts triggered at critical nodes.

At the same time, the platform integrated seamlessly with their PostgreSQL database, enabling data to be accessed both within NocoBase and through external SQL queries or BI tools, preserving the existing data ecosystem.

In their typical use cases, NocoBase served as the core infrastructure for the AI multimodal R&D platform:

  • As an AI scheduling and management hub, it centralized the orchestration of different compute and model pipelines, eliminating fragmented processes.
  • On the content generation side, it supported large-scale image generation and management, while laying the groundwork for future multimodal expansion—truly bridging text, image, and other content types.

Task-17kdnf.png

Management.png

This not only shortened development cycles but also improved the overall R&D experience.

Team members were especially impressed by the platform’s ease of use and scalability:

“As back-end developers, we don’t have to worry about front-end interfaces anymore. NocoBase covers almost all of our business design needs. With workflows handling complex logic, the efficiency gains are remarkable.”

Although the platform is still in its R&D phase, it has already been deployed internally and is demonstrating significant value.

Outlook

For Wasu Media, this is just the beginning.

By unifying AI workflows and data pipelines under one roof, they’re setting the stage for scalable, multimodal innovation — powered by NocoBase.

And this story shows how no-code can accelerate even the most advanced AI R&D.

More Customer Stories:

× View Image