Pipedream alternative: FlowRunner vs Pipedream

TL;DR

• Pipedream Workflows and String shut down on March 31, 2027. All Workflows data, including the credentials and OAuth tokens for connected accounts, is deleted by April 30, 2027. Signups and subscriptions are already closed.
• Export every project now. The ZIP has one JSON file per workflow with its triggers, steps, and code. It leaves out secrets, environment variables, data stores, File Store files, and event history.
• FlowRunner runs the same work: webhooks, schedules, and app triggers from 2,100+ integrations, JavaScript code steps, and your own blocks with npm packages. Approvals reach a person on email, Slack, WhatsApp, phone, or a Form and wait up to a year on Professional and above, and a complete run counts as one execution.
• FlowRunner Prompt to Flow converts the export one workflow at a time, code steps included. The migration guide has the steps.

What closes, and when

Pipedream’s workflows docs put it plainly: “Pipedream Workflows and String will shut down on March 31, 2027.” Here is the timeline.

• Now. Workflows is in maintenance mode. No new signups, no new subscriptions, no new features. Existing workflows keep running.
• March 31, 2027. Workflows and String stop. Pipedream cancels subscriptions and issues a prorated refund.
• April 30, 2027. All Workflows data is permanently deleted, including the credentials and OAuth tokens for your connected accounts.
• Not affected. Pipedream Connect, MCP Servers, and the API Proxy.

The export is the only thing you take with you, and it carries the definitions, not the state. Event history, data store records, File Store files, and environment variables stay behind. Copy the data store records and save a few real events from each workflow while it still runs. The events are what you’ll test the new flow against.

Who this comparison is for

You run workflows on Pipedream and need somewhere to put them before March 31, 2027. You probably wrote code steps in Node.js, maybe some Python, maybe a few components or event sources of your own. Some of your workflows may pause with $.flow.suspend and wait for someone to click a link.

Start with what these two share: event-driven workflows that start from a webhook, a schedule, or an app event, run prebuilt actions, and drop into code where a prebuilt action runs out. FlowRunner does all of that. On top of it, FlowRunner runs AI agents that call your flows as tools, approvals that find the person and wait for them, and a record of every run that the whole team can open. Read the two side by side and the shape is plain.

What FlowRunner is good at

FlowRunner is built around agents that know when to stop and ask a person. Human-in-the-loop is an execution pattern where AI agents pause autonomously, assemble the relevant context and the decision choices available, route to a human via their preferred channel, and resume the moment the human responds. A flow can do the same without an agent. The rest of the platform is there to make that pause dependable.

• Approvals that reach a person. The run sends the request on email, Slack, WhatsApp, phone, or a Form, waits up to 30 days on Free, Starter, and Growth and up to a year on Professional and above, and picks up where it stopped. An AI agent can call that approval as one of its tools and suspend until the answer comes back.
• One execution per run. An execution is one complete run, start to finish, however many steps it has and however long it waits.
• Code where code fits. Custom Cloud Code is a JavaScript step with fetch, crypto, Buffer, timers, and URL, and it can also be attached to an AI agent as a tool. When you need npm packages, a custom extension is one JavaScript module deployed with the FlowRunner CLI, with its node_modules shipped in the deploy. Extensions add actions, polling and webhook triggers, OAuth, and a settings form, and every deploy is a version you can roll back. npx flowrunner init-claude installs Claude Code agents that write the extension and its tests.
• AI agents on your own keys. The AI Agent block uses your own provider key with no markup (BYOK). Its tools can be integration actions, your other flows, MCP servers you register, and Knowledge Bases.
• A deep catalog. 2,100+ integrations and 56,000+ callable actions (as of September 2026), and every action carries the metadata an agent needs to call it as a tool.
• A record of every run. Each run is stored block by block, with each block’s input and output. A failing block can route to its own Handle Error path, and the HTTP Request block retries on the status codes you list.
• Room for everyone. Unlimited users and unlimited workflows on every plan, Free included.

What Pipedream is good at

Pipedream Workflows was built for developers who would rather write a step than configure one. A workflow is a trigger and a list of steps, and any step can be code. Developers publish their own components and event sources, control the run from code with $.flow.delay, $.flow.suspend, $.flow.exit, and $.respond, and keep projects in a Git repository with GitHub Sync.

Pipedream’s own focus has moved to Connect, its API for putting integrations inside other products, along with MCP Servers and the API Proxy. Those keep running after March 31, 2027.

Where FlowRunner is stronger

• Approvals that find the person and wait. $.flow.suspend pauses a workflow until something calls its resume or cancel URL. Delivering that URL to a person is up to you, the suspend auto-cancels after 24 hours unless you set a longer timeout, and each reawakening counts as a new credit. A FlowRunner approval sends the request itself, on email, Slack, WhatsApp, phone, or a Form, waits up to 30 days on Free, Starter, and Growth or up to a year on Professional and above, and the run still counts once.
• A meter that doesn’t move with compute. Pipedream charges a credit per 30 seconds of compute at 256 MB, raises the burn when you give a workflow more memory, and meters AI tokens separately. FlowRunner counts one execution per complete run. AI runs on your own provider keys, billed by the provider, with no markup.
• Long-running work. A Pipedream execution stops at 300 seconds on free tiers and 750 seconds on paid ones. A FlowRunner run can compute for up to 1 hour on Free, Starter, and Growth, 4 hours on Professional, 12 hours on Business, and without a cap on Enterprise, and it can wait at an approval or a Wait block for days and then carry on. Waiting doesn’t count toward that time.
• Unlimited workflows on every plan. Pipedream Free allowed 3 active workflows and Basic allowed 10. FlowRunner has no workflow limit on any plan, Free included, and no user limit either.
• Depth per integration. Pipedream lists 3,000+ APIs and 10,000+ prebuilt tools and triggers. FlowRunner has 2,100+ integrations with 56,000+ callable actions, and each action is ready for an agent to call as a tool.
• Compliance at a published price. Pipedream signs a BAA for Business customers only and provides its SOC 2 report on request. FlowRunner includes audit trails and RBAC from Professional at $299 a month, and signs a BAA on Business at $999 a month, which also adds SSO/SAML.
• Agents that keep going. String closes on the same day as Workflows. FlowRunner’s AI Agent block takes your prompt as it is, on your own key, and can call your flows, including an approval flow, as tools.

Where Pipedream is stronger

• Code in four languages. Pipedream code steps run Node.js, Python, Go, and Bash, and a Node.js step can import an npm package right in the step. FlowRunner runs JavaScript only. JavaScript steps move into Custom Cloud Code nearly as written, and steps with packages become a custom extension that ships its dependencies. Python, Go, and Bash steps get rewritten in JavaScript, which FlowRunner Prompt to Flow does as part of each conversion. A long Python step you’d rather keep can run as a small service of your own that the flow calls. After March 31, 2027, Pipedream isn’t running any of that code either way.
• Workflows as files in Git. GitHub Sync keeps each workflow as a workflow.yaml plus one file per code step, in your own repository. FlowRunner keeps versions of each flow in the workspace and exports any version as a JSON file. Your GitHub Sync files don’t go to waste: they are one of the two inputs FlowRunner Prompt to Flow converts from.

Starting free

Pipedream’s Free plan gave you up to 100 credits a month, 3 active workflows, and 3 connected accounts, and it’s closed to new signups. A credit covers 30 seconds of compute, so a multi-step workflow could use several per run.

FlowRunner’s Free plan is $0, permanent, and needs no card:

• 100 executions a month, where one execution is one whole run
• one run at a time, with 24 hours of log history
• unlimited users and unlimited workflows
• every integration, AI agents on your own keys, and human-in-the-loop

Every new account also gets 14 days of Professional on its first workspace, with audit trails and RBAC. If no card is added when the trial ends, the workspace moves to Free and the flows keep running. Above Free, Starter is $5 a month for 300 executions or $15 for 3,000, and Growth starts at $45 for 12,000.

Feature comparison

Dimension	FlowRunner	Pipedream Workflows
Status	Active	Shuts down March 31, 2027; data deleted by April 30, 2027
Billing unit	Execution: one complete run	Credit: 30 s of compute at 256 MB; AI tokens metered separately
A run that waits two days for approval	One execution	Suspended time not billed; each reawakening is a new credit
Free plan	$0, 100 executions a month, no card, open to new accounts	Up to 100 credits a month, 3 active workflows; closed to new signups
Paid entry price	$5 a month Starter (300 executions); $15 for 3,000	Basic $29 a month; closed to new subscriptions
Workflows per plan	Unlimited on every plan	3 active on Free, 10 on Basic, unlimited on Advanced
Triggers	External Callback (webhook), schedules, integration triggers, API call, another flow	HTTP/webhook, schedule, email, app event sources
Code steps	JavaScript in Custom Cloud Code, with fetch; also an agent tool	Node.js, Python, Go, Bash
Packages and your own blocks	Custom extensions: npm packages, actions, polling and webhook triggers, OAuth, versions with rollback	npm imports in Node.js steps; components and event sources
Human approval	Reaches the person on email, Slack, WhatsApp, phone, or a Form	$.flow.suspend with resume and cancel URLs you deliver yourself
Longest approval wait	30 days on Free, Starter, Growth; 1 year on Professional and above	Auto-cancels after 24 hours by default; configurable
Execution time limit	Active runtime of 1 hour (Free to Growth), 4 hours (Professional), 12 hours (Business), unlimited (Enterprise); waiting doesn’t count	300 s on free tiers, 750 s on paid
AI agents	AI Agent block on your own keys; tools include actions, flows, MCP servers, Knowledge Bases	String (closes with Workflows)
MCP	Agents use tools from registered MCP servers; FlowRunner Prompt to Flow lets AI tools build and run flows	Pipedream MCP Servers (not affected by the shutdown)
Catalog	2,100+ integrations, 56,000+ callable actions	3,000+ APIs, 10,000+ prebuilt tools and triggers
Errors and retries	Handle Error path per block; Retry Policy on HTTP Request	Error notifications; auto-retry (up to 8 times over 10 hours) on Advanced
Run record	Every run stored block by block, with inputs and outputs	Event history per workflow
Versions	LIVE, READY, and NOT READY versions per flow; any version exports as JSON	GitHub Sync
Compliance	Audit trails and RBAC from Professional ($299); a HIPAA BAA and SSO/SAML from Business ($999)	SOC 2 Type 2 report on request; BAA for Business customers only
Where it runs	FlowRunner cloud; on-prem on Enterprise	Pipedream cloud

Decision framework

Pick FlowRunner if:

• Your Pipedream workflows need to keep running after March 31, 2027.
• One of them pauses for a person, and you want the request to reach that person and wait days, not 24 hours.
• Your code is mostly JavaScript, or short enough to rewrite, and the packages it needs can ship in a custom extension.
• You want one execution per run, AI on your own keys, and unlimited workflows and users.
• You handle health data and need audit trails, RBAC, and a BAA at a published price.

Pick Pipedream if:

• You only use Pipedream Connect, MCP Servers, or the API Proxy, which the shutdown doesn’t touch.
• The workflow is something you can retire on March 31, 2027.

Ready to move?

You don’t rebuild your workflows by hand. The Pipedream migration guide shows how FlowRunner Prompt to Flow converts one workflow at a time from the export or your GitHub Sync files, and what’s left for you to connect and test. For a bigger migration, contact us.