Flowise

Flowise is an open-source platform for building AI agents and LLM workflows with visual tools. It is best suited to developers and teams comfortable running or forking software; its breadth is useful, but its wind-down makes it a risky choice for new production work.

Overview

Assistant, Chatflow and Agentflow builders address assistants, chatbots, single- and multi-agent systems, and workflow orchestration. That range lets teams use one platform for everything from a focused chatbot to coordinated workflows. The trade-off is decisive: feature development stopped on July 29, 2026, repository archival is scheduled for August 10, and end of life for August 31, 2026.

The source remains on GitHub under the Apache 2.0 license, and Flowise encourages teams to fork it. That may suit organizations able to take responsibility for ongoing maintenance, but it does not preserve active upstream development or official core-team support.

Key features

Execution logs, visual debugging, evaluations, human review and RAG indexing pipelines complement the visual builders. In practice, teams can inspect runs, assess outputs and include review steps rather than only connecting workflow components. Support for knowledge grounding and multi-agent systems broadens the kinds of applications it can accommodate.

Flowise connects to more than 100 sources, tools, vector databases and memories, including LangChain, LlamaIndex and Zapier Zaps. APIs, an embedded chat widget, and TypeScript and Python SDKs give developers several ways to integrate their work. Security controls include RBAC, SSO, encrypted credentials, secret managers, rate limiting and restricted domains. For production, its guidance recommends queue mode, PostgreSQL at scale, AWS S3 storage and AWS Secret Manager for credential key management.

Deployment can be cloud or on-premises, with npm, Docker and self-hosting options. The platform lists Linux, macOS and Windows as well as web, API and self-hosted availability. These options are useful for teams that need control over deployment, but a new installation should account for deprecated npm packages and Docker images after repository archival.

Pricing

The Free plan costs nothing and includes two Flows and Assistants, 100 predictions per month, 5MB of storage, evaluations and metrics, custom embedded-chatbot branding, and community support. It is suitable for small experiments, but the flow, prediction and storage caps constrain ongoing use.

Starter costs $35 a month. It adds unlimited Flows and Assistants, 10,000 predictions per month and 1GB of storage while retaining community support. It fits modest workloads that outgrow the free quotas, though it does not include the higher limits or administrative controls in Pro.

Pro costs $65 a month and raises the allowance to 50,000 predictions and 10GB of storage, with unlimited workspaces, admin roles and permissions, and priority support. It includes five users, with additional users at 15 USD per user per month. That makes it the more capable option for teams needing separate workspaces and access controls, but the user charge adds to the monthly cost. The free plan, paid tiers and a free trial are offered; given the announced wind-down, committing to a paid plan for long-term use deserves particular caution.

Platforms

Flowise supports web and API access, Linux, macOS and Windows, plus cloud, on-premises and self-hosted deployment. This combination suits teams that want to connect workflows programmatically or run them on infrastructure they manage.

Who it's for

Flowise is most appropriate for developers evaluating visual agent and LLM workflow construction, or organizations prepared to maintain a fork after upstream development ends. Its builders, integrations, debugging and deployment choices make it capable for experimentation and internal projects. It is a poor fit for buyers who need an actively maintained vendor-backed platform for a new long-lived production service.

Pros and cons

• Pros: Three visual builders cover assistants, chatbots and single- or multi-agent workflows, giving teams flexibility within one platform.
• Pros: More than 100 integrations, developer interfaces, evaluation tools and security controls support connected, inspectable workflows.
• Pros: Apache 2.0 source and self-hosting give capable teams the option to continue independently.
• Cons: Feature development has ceased and end of life is scheduled for August 31, 2026, undermining its suitability as a new production foundation.
• Cons: Official core-team presence in Discord and GitHub concluded on August 31, 2026, limiting the support path as the project winds down.
• Cons: The free tier's two-flow, 100-prediction and 5MB limits are narrow for sustained use; higher quotas require paid plans.

Alternatives

For readers comparing broader AI Agent Platforms or focused Retrieval-Augmented Generation Tools, the right choice depends on whether agent orchestration or retrieval is the priority. AnythingLLM is another freemium option spanning web, desktop, mobile, API and self-hosted use. RAGFlow offers a free plan with five apps, one team member, 0.1 GB dataset storage and 500 monthly credits, making those caps relevant to smaller evaluations. I3K RAG Enterprise has a free, self-hosted Community plan under AGPL-3.0 for Linux or Windows. Dify offers a free Sandbox with 200 message credits, one workspace, one member and five apps. Haystack Enterprise Platform has a free Studio plan with one workspace, one user and 100 pipeline hours, plus a free trial. OpenRAG provides an AGPL-3.0 self-hosted plan with no features held back. Ragen offers self-hosted use without license fees, with users paying for infrastructure and models. Google Cloud Agent Evaluation is a paid service with a free trial and usage-based charges for evaluation metrics.

Verdict

Choose Flowise if you need visual construction for agents and LLM workflows and can accept responsibility for a fork or a short-lived project. Its combination of builders, integrations, evaluation tools and deployment choices is compelling, but the end-of-life schedule and concluded core-team support are strong reasons to look elsewhere for a production system expected to last.