Repolode
Submit
dataelement

dataelement/bisheng

An open-source LLM application devops platform for enterprises, covering workflows, RAG, agents, model management, evaluation and fine-tuning.

View on GitHubWebsite
Stars
12,068
Forks
1,981
Open issues
19
Last commit
2 hours ago

What is it?

BISHENG is an open-source LLM application platform aimed at enterprise deployments. It lets teams build AI applications through a visual workflow orchestrator that supports loops, parallel branches, batch runs and pausing for human input, in a single flow that can also handle multi-turn conversation.

The platform includes a task mode in which an agent breaks a goal into steps, runs sub-agents in parallel, and produces reports, PPT, Excel or Word files. A knowledge space module provides team knowledge bases with multi-level folders, file-level permissions and AI Q&A that cites its sources. Document parsing models for OCR, table recognition and layout analysis can be deployed privately.

For administration, BISHENG provides department and user management, RBAC and ReBAC permissions, approval flows, audit logs and an Open API with API keys. Self-hosting uses Docker Compose, and the README notes a recommended setup of 18 cores and 48 GB RAM, with Elasticsearch, Milvus and OnlyOffice installed by default.

Key features

  • 01Visual workflow orchestration with loops, parallel branches and batch runs
  • 02Task mode agent that plans steps, runs sub-agents and exports reports or Office files
  • 03Knowledge space with multi-level folders, file-level permissions and cited AI Q&A
  • 04Document parsing models for OCR, handwriting, tables, layout and seals
  • 05Enterprise controls including RBAC/ReBAC, approvals, audit logs and SSO/LDAP
  • 06Open API with service accounts and API keys

Use cases

  • Building internal AI assistants that answer questions over company documents
  • Automating document review or fixed-layout report generation with human approval steps
  • Deploying a private LLM platform that must run on domestic IT stacks such as Dameng DM8
  • Digitizing scanned documents with privately deployed OCR and table recognition

Quick start

  1. 1

    Check hardware and Docker prerequisites (4+ cores, 16 GB RAM, Docker 19.03.9+, Docker Compose 1.25.1+)

  2. 2

    Clone the repository and enter the docker directory

    git clone https://github.com/dataelement/bisheng.git && cd bisheng/docker
  3. 3

    Start BISHENG with Docker Compose

    docker compose -f docker-compose.yml -p bisheng up -d
  4. 4

    Open http://IP:3001 in a browser and register; the first registered user becomes admin

Things to know

  • Recommended hardware is high (18 cores, 48 GB RAM), and it installs ES, Milvus and Onlyoffice by default.
  • Some features such as dashboards and DSH Desktop access are only in the commercial edition.

Topics

#llm#rag#agent#workflow#enterprise#ocr#llmops#knowledge-base#ai#chatbot

FAQ

What is bisheng?
An open-source LLM application devops platform for enterprises, covering workflows, RAG, agents, model management, evaluation and fine-tuning. BISHENG is an open-source LLM application platform aimed at enterprise deployments. It lets teams build AI applications through a visual workflow orchestrator that supports loops, parallel branches, batch runs and pausing for human input, in a single flow that can also handle multi-turn conversation.
Is bisheng free to use? What license does it have?
bisheng is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is bisheng still maintained?
Actively maintained. The last commit to bisheng was on Oct 21, 2026.
What can I use bisheng for?
Building internal AI assistants that answer questions over company documents Automating document review or fixed-layout report generation with human approval steps Deploying a private LLM platform that must run on domestic IT stacks such as Dameng DM8 Digitizing scanned documents with privately deployed OCR and table recognition
How do I install bisheng?
Check hardware and Docker prerequisites (4+ cores, 16 GB RAM, Docker 19.03.9+, Docker Compose 1.25.1+) Clone the repository and enter the docker directory: git clone https://github.com/dataelement/bisheng.git && cd bisheng/docker Start BISHENG with Docker Compose: docker compose -f docker-compose.yml -p bisheng up -d Open http://IP:3001 in a browser and register; the first registered user becomes admin
Where can I get bisheng?
The source code and installation instructions are on GitHub: https://github.com/dataelement/bisheng

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects