Dify review: self-hostable agent workflows with a human decision step

Contents

1. The human decision step
2. Hosting and licence
3. How it is billed
4. Strengths and limits
5. Pricing
6. Questions to ask before building on Dify
7. Sources

Type	Agentic workflow platform
Pricing	Published
Website	dify.ai

Dify is a platform for agentic workflows: a visual studio, a knowledge pipeline for assistants built on company documents, and marketplace tools and MCP. It runs in Dify's cloud, in a private cloud or self-hosted⁠1Dify: the platform for production-ready agentic workflowsLangGenius (Dify).

Best for: Technical teams building agent workflows and knowledge-based assistants they can self-host, with a human decision step that can be answered by email.

Not the right fit for: Operations staff configuring agents alone, or companies that want to offer the platform to their own customers.

The human decision step

Dify's Human Input node pauses a workflow and presents a request form with custom decision buttons, each leading to a different path. Requests can be delivered in a web app or by email, and email recipients need no Dify account. The request closes after the first response; by default it times out after three days and follows a timeout branch. The web-app method is not available in workflows started by a trigger⁠2Human Input nodeDify Docs.

Hosting and licence

The Community edition is free and self-hosted with Docker, limited to a single workspace; Enterprise adds SSO, role-based access, audit logs and a Helm chart for Kubernetes⁠31Dify pricingDifyDify: the platform for production-ready agentic workflowsLangGenius (Dify). Dify's licence is a modified Apache 2.0 that forbids operating a multi-tenant environment from the source code without written authorisation and forbids removing its logo or copyright information from the console⁠4LICENSE (Dify Open Source License)langgenius/dify on GitHub.

How it is billed

Cloud plans are priced per workspace per year: a free Sandbox with 200 message credits, Professional at $590 a year with 5,000 message credits a month and three members, and Team at $1,590 a year with 10,000 credits and 50 members⁠3Dify pricingDify.

Strengths and limits

Strengths

• The Human Input node lets a person decide a workflow’s next path with custom buttons; email reviewers need no Dify account.
• Strong on knowledge: a pipeline for building assistants on company documents.
• Can be self-hosted, and Enterprise adds SSO, RBAC, audit logs and a Helm chart.

Limits

• Its modified Apache 2.0 licence forbids operating a multi-tenant service without written authorisation.
• The Community edition is a single workspace; SSO, RBAC and audit logs are Enterprise features.
• Human Input requests time out (three days by default), and the web-app method is unavailable in trigger-started workflows.
• Built for builders: designing reliable workflows assumes technical skill.

Choose another option if: The agents will be set up by operations staff rather than developers. Lindy, Relevance AI or OpenLazarus are closer to that need.

Pricing

Published: Sandbox free (200 message credits); Professional $590 and Team $1,590 per workspace a year. Enterprise on request.

Questions to ask before building on Dify

• Who will design and maintain the workflows?
• Which decisions should pause for a person, and what should happen on timeout?
• Will we self-host, and does our use fit the licence?
• Do we need Enterprise for audit logs and roles?

See how Dify compares with the other agent builders in the main ranking.

Sources

1. Dify: the platform for production-ready agentic workflows. LangGenius (Dify). Accessed 8 Nov 2026.
2. Human Input node. Dify Docs. Accessed 8 Nov 2026.
3. Dify pricing. Dify. Accessed 8 Nov 2026.
4. LICENSE (Dify Open Source License). langgenius/dify on GitHub. Accessed 8 Nov 2026.

Further reading

• Buyer's guide · Updated quarterlyThe best no-code AI agent builders for company work in 2026An analysis of seven platforms for building AI agents without code, judged on the controls a company needs before an agent changes a real order, invoice or customer record.
• Guide · ControlAI agent permissions and approvals: how to let agents act safelyAn agent's permissions are a job description written in advance. The platforms differ less in whether they offer approvals than in where those approvals apply and what record they leave.
• Guide · CostWhat no-code AI agents cost: credits, actions, executions and the monthly billEvery platform meters agent work in its own unit. Converting those units into the cost of one task is the only way to compare them, and the only way to avoid the cost overruns Gartner expects to end many projects.