dataelement/bisheng

★ 11,363TypeScriptAudience · pm founderComplexity · 4/5Setup · hard

Mindmap

mindmap
  root((repo))
    What it does
      AI workflow builder
      Document parsing
      Enterprise agents
    Features
      Visual canvas
      Human in loop
      Role based access
    Use cases
      Contract review
      Document extraction
      Team AI management
    Setup
      Docker Compose
      16GB RAM minimum
      Browser interface

mindmap root((repo)) What it does AI workflow builder Document parsing Enterprise agents Features Visual canvas Human in loop Role based access Use cases Contract review Document extraction Team AI management Setup Docker Compose 16GB RAM minimum Browser interface

Click or tap to explore — scroll the page freely

Things people build with this

USE CASE 1

Build a visual AI workflow that reads uploaded contract PDFs and automatically extracts key clauses and dates.

USE CASE 2

Deploy an expert AI agent trained on your company's business rules to handle internal queries as a domain specialist.

USE CASE 3

Process large volumes of scanned forms or printed documents using the built-in high-precision document parser.

USE CASE 4

Manage AI usage across teams with role-based access control, SSO, and per-team usage tracking.

Tech stack

TypeScriptPythonDockerDocker Compose

Getting it running

Difficulty · hard Time to first run · 1h+

Requires Docker Compose with a minimum of 4 CPU cores and 16 GB RAM to run the full platform.

In plain English

BISHENG is an open-source platform built for companies that want to create and manage AI-powered applications at scale. The name comes from Bi Sheng, the inventor of movable type printing in ancient China, and the project carries the same ambition: spreading information and intelligence more widely. It is developed by a Chinese team and has been adopted by large enterprises and Fortune 500 companies according to the README. The core of the platform is a workflow builder that lets teams design multi-step AI processes visually, like drawing a flowchart. Unlike similar tools, BISHENG supports loops, parallel execution, batch processing, and conditional branching all within the same visual canvas. Users can also pause a running workflow to review or correct the AI's output before it continues, which the README describes as "human in the loop." For AI interactions, BISHENG includes an agent called Linsight that is designed to carry expert knowledge about a domain. The idea is that business rules, workflows, and industry-specific preferences can be embedded into how the agent reasons, so it behaves less like a generic assistant and more like a trained specialist. The platform also includes high-precision document parsing built on five years of training data. It can read printed text, handwritten text, unusual characters, tables, and complex page layouts. This is useful for enterprise tasks like reviewing contracts, extracting data from forms, or processing call records. Enterprise management features include role-based access control, single sign-on (SSO), usage tracking by team, security review, and high-availability deployment options. The platform also covers model fine-tuning and dataset management for teams that want to train or adapt their own models. Installation runs through Docker Compose and requires at least 4 CPU cores and 16 GB of RAM. After startup, the interface is available in a browser, and the first registered user becomes the system administrator.

Copy-paste prompts

Prompt 1

Using BISHENG, design a workflow that reads an uploaded contract PDF, extracts key dates and party names, and outputs a structured JSON summary.

Prompt 2

How do I configure a Linsight agent in BISHENG to follow our company's specific approval rules and business terminology?

Prompt 3

Walk me through setting up BISHENG with Docker Compose, creating the admin account, and inviting a team member with editor access.

Prompt 4

Show me how to add a human-in-the-loop review step to a BISHENG workflow so a manager must approve the AI output before it continues.

Prompt 5

What BISHENG features handle SSO integration and role-based access control for a 500-person enterprise deployment?

Open on GitHub → Explain another repo

← dataelement on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.