MMake

OpenHuman is an open-source, local-first personal AI assistant built around persistent memory, research, agent orchestration, and connected tools. It imports information from services such as Gmail, Notion, GitHub, and Slack, stores compressed Markdown-based memory in SQLite, and supports goals, todos, messaging channels, web search, browsing, coding, voice, and media generation.

Its workflow feature lets an agent propose automation graphs that users review on a visual canvas before saving. Workflows can be durable, trigger-driven, approval-gated, and activated by schedules, webhooks, or channel events; checkpointed runs can pause, survive restarts, and resume. The project provides desktop installers and terminal installation options, with local data storage, optional local models, and a privacy mode that prevents inference from leaving the machine.

Conductor is an open-source workflow engine for orchestrating microservices, AI agents, and durable workflows. It persists every step, supports retries, timeouts, dynamic forks, loops, sub-workflows, versioning, replay, and recovery from failures. Workers can be written in Java, Python, Go, JavaScript, C#, Ruby, or Rust, while workflows use a declarative JSON model.

The product includes a built-in visual UI for designing, running, observing, and debugging workflows, with task input, output, timing, and retry history inspection. It also documents native support for 14+ LLM providers, MCP and function calling, human approvals, vector databases for RAG, and runtime-generated workflow definitions.

Conductor can run locally through its CLI, Docker, or a JVM-based server. It is licensed under Apache 2.0 and supports self-hosted deployments with multiple persistence backends and message-brok不

Sim is a collaborative workspace for building, deploying, and managing AI agents and workflows. Users can create agents visually, conversationally, or with code; connect more than 1,000 integrations and major language models; ingest files, knowledge bases, and structured table data; and monitor runs, logs, schedules, and workflow activity.

The workspace includes tables, files, and searchable knowledge bases for teams and agents. It can be used through the hosted sim.ai service or deployed locally from the repository. The README documents local development, Docker Compose, Kubernetes via Helm, and support for local models through Ollama and vLLM.

Skyvern automates browser-based workflows with large language models and computer vision. Users can define tasks with prompts, chain browser tasks into workflows, navigate unfamiliar websites, fill forms, extract structured data, validate page state, download files, and perform actions through AI-powered Playwright commands.

The repository documents a packaged web UI, Python and TypeScript SDKs, a cloud service, and local deployment. Users can install it with pip, run a local server and UI, or start the complete stack with Docker Compose and PostgreSQL. Workflow blocks include browser actions, data extraction, loops, file parsing, email sending, HTTP requests, custom code, and block storage uploads.

AstronRPA is an enterprise-grade robotic process automation desktop application for building workflows that automate Windows desktop software, web pages, browsers, office tools, financial systems, and ERP applications. Its visual designer supports low-code and no-code process creation, debugging, and custom component extensions.

The suite includes more than 300 pre-built capabilities for UI operations, data processing, system interactions, browser automation, files, email, documents, APIs, and AI services. Workflows can run directly, on schedules, through scheduling modes, API calls, or MCP services, with support for monitoring, permissions, team sharing, and collaboration.

The repository documents Windows client deployment and a Docker-based server deployment. The desktop architecture uses Vue, TypeScript, and Electron, with Java Spring Boot and Python FastAPI services. It is released

SuperPlane is an open-source automation engine and control plane for AI-driven engineering. It lets teams define git-backed apps and canvases that combine workflow graphs, custom console interfaces, app-scoped memory, and deterministic execution.

Workflows can span Git, CI/CD, infrastructure, observability, incident management, notifications, and AI tools. They support webhooks, schedules, tool events, approvals, policy checks, human-in-the-loop steps, durable runs, resumable failed steps, and concurrent execution. Documented examples include preview environments, gated deployments, progressive delivery, release trains, and incident triage.

The core engine can be self-hosted, with a documented Docker demo-container path, or used through SuperPlane Cloud. The README also documents operational dashboards, built-in per-app agents, CLI access, RBAC, and numerous integrations.

OpenFn Lightning is an open-source workflow automation platform for automating critical business processes and integrating information systems. It is used by NGOs and governments for workflows ranging from last-mile services to national-level reporting.

Users can visually build workflows, monitor activity, filter and search runs, configure failure alerts, review project digests, and manage users and project access through roles and permissions. A CLI is also documented for building, editing, and deploying projects from a code editor.

Lightning is available as OpenFn-hosted software, a public sandbox, or a self-hosted application. The repository documents Docker and Docker Compose deployment with PostgreSQL, external infrastructure deployment including Kubernetes, and local development through Elixir and Phoenix.

LAYRA is an open-source visual AI-agent engine combining visual document understanding, multimodal retrieval-augmented generation, and workflow orchestration. Users can upload and query documents, preserve layout and graphical elements, and create workflows with branching, nested loops, conditions, human approval steps, MCP integration, chat memory, and multimodal inputs and outputs.

The platform provides a Next.js web interface and FastAPI backend, with workflow debugging, streamed execution, reusable components, sandboxed Python execution, and integrations with services including Redis, MySQL, MongoDB, Kafka, MinIO, and Milvus. It is deployed under user control with Docker Compose; local embedding models can use NVIDIA GPUs, while Jina Embeddings v4 is available through a cloud API.

Langflow is an open-source platform for building and deploying AI-powered agents and workflows. Its visual builder supports iterative authoring, while source-code access allows developers to customize components with Python. An interactive playground provides step-by-step testing and refinement, and the platform documents multi-agent orchestration, conversation management, retrieval, and integrations with major LLMs, vector databases, observability tools, and AI tools.

Users can deploy flows as APIs, export them as JSON for Python applications, or expose them as MCP servers. Langflow Desktop is available for Windows and macOS, and the project can run locally through its Python package or in Docker. The repository also documents deployment to major cloud environments.

Trigger.dev is an open-source platform for building AI workflows and agents in TypeScript. Developers define long-running background tasks in their codebase and deploy them with durable execution, retries, queues, idempotency, checkpointing, and elastic scaling.

The platform supports cron schedules, batch triggering, structured inputs and outputs, human-in-the-loop waits, realtime updates and streaming, concurrency controls, multiple environments, versioning, logging, tracing, alerts, and run-level observability. It can run browsers, Python scripts, and FFmpeg through build extensions.

Users can run tasks through Trigger.dev’s cloud or self-host the complete product. The README documents Docker Compose and Kubernetes deployment paths, as well as a web app for creating projects and viewing task runs.

RocketRide is an open-source AI development environment and pipeline builder for composing, debugging, observing, and deploying AI and machine-learning workflows. Its VS Code extension provides a visual canvas for connecting nodes, while pipelines use portable JSON and can run as standalone processes or through Python and TypeScript applications.

The documented feature set includes more than 100 nodes covering LLM providers, vector databases, OCR, NER, anonymization, embeddings, and other data-processing tasks. It supports multi-agent workflows through CrewAI and LangChain, MCP and SDK integrations, execution tracing, token and latency monitoring, and a multithreaded C++ runtime.

RocketRide can run through RocketRide Cloud or under user control as a local process, Docker deployment, on-premises installation, bare metal service, or clustered deployment with Helm. The repository is MIT-ไ

Loonflow is an open-source process automation platform built with Django and React. Users can design business processes with drag-and-drop nodes, conditional branches, parallel tasks, hooks, forms, validation, previews, and multiple process versions. It supports ticket types for IT operations, HR approvals, financial reimbursements, and customer service, with conditional routing and automatic assignee assignment.

The platform also provides plugin and REST API extension points, granular permissions, audit logs, authentication through Microsoft OIDC and WeCom QR login, optional multi-tenant support, and an MCP ticket server. The repository documents deployment through Docker Compose, including local configuration and startup instructions, alongside a separately offered managed SaaS edition.

RuoYi AI is an enterprise AI assistant platform for building AI agents and coordinating multi-agent systems. Its documented modules include multi-provider model management, local RAG with vector databases, document parsing, MCP integration, extensible skills, and a LangChain4j-based agent framework with Supervisor orchestration.

The platform includes a visual workflow designer with drag-and-drop node orchestration and SSE streaming execution. Documented nodes include model calls, email sending, and manual review. It provides user and admin frontends, a backend service, monitoring, and WebSocket communication. The MIT-licensed project supports Docker Compose deployment, including bundled MySQL, Redis, Weaviate, and MinIO services, or step-by-step source builds.

xyOps is an open-source workflow automation and server monitoring platform for developers and operations teams. It schedules jobs across server fleets and provides a graphical workflow editor for connecting events, triggers, actions, and monitors into operational pipelines.

The platform combines real-time monitoring, customizable alerts, server snapshots, ticket creation, logs, execution history, and incident-response context. It is designed to connect detection, automation, and resolution in one system, with support for deployments ranging from small installations to large server fleets.

xyOps is BSD-3-Clause licensed and documents self-hosting through its hosting guide. Local development uses Node.js, npm, and the included build and debug scripts; the README also describes free self-hosted use and paid support tiers.

Nyno is an open-source workflow builder and execution engine aimed at EU-AI applications. Users define workflows in a YAML-like format, including AI steps, prompts, context such as chat history, and additional processing steps. The README documents building and testing workflows with a Run Workflow action, then exporting them as `.nyno` files.

The product can run locally through Docker, exposing a service on port 9057 with mounted PostgreSQL data and enabled-workflow directories. Workflows can be enabled by copying files into the configured folder and called from a frontend through an HTTP API. The documentation also describes workflow-user creation, authentication, API-key masking, and deployment-related secret configuration.

Dograh is an open-source voice AI platform for creating production voice agents. Its visual workflow builder supports start and agent nodes, instructions, tools, transitions, end-call outcomes, knowledge bases, webhooks, tool calling, QA, browser audio testing, and text-based test conversations.

The platform connects telephony providers including Twilio, Vonage, Telnyx, Plivo, Vobiz, Cloudonix, and Asterisk ARI. Users can bring their own LLM, TTS, STT, and telephony providers, use MCP clients to inspect or edit agent workflows, and create agents or outbound calls with Python and Node SDKs.

Dograh can run locally or on a remote server using Docker, with bundled MinIO or S3-compatible storage. A managed cloud version is also offered, while the repository documents self-hosted deployment and a BSD 2-Clause license.