Pipedream

Name: Pipedream
Brand: Pipedream
Price: 1002945 USD

Code-level workflow automation across 3,000+ APIs, billed by compute rather than by task

Pipedream is a developer-first integration and workflow automation platform. You build event-driven workflows in a browser, mixing pre-built triggers and actions for more than 3,000 apps with arbitrary Node.js, Python, Go, or Bash steps, and Pipedream runs them on managed serverless workers with authentication handled for you. It is priced by credits of compute time rather than per task, starts free, and also ships Connect, an SDK and MCP layer that lets a product embed those same integrations for its own end users.

Visit website

Overview

Most automation tools ask you to express your logic as a chain of configured boxes, and they charge you every time a box runs. Pipedream started from the opposite premise: give a developer a hosted runtime with the boilerplate removed, then attach a very large library of pre-authenticated app connectors to it. A step in a Pipedream workflow can be a verified Slack action picked from a dropdown, or it can be forty lines of Python that import a PyPI package and read a secret from an environment variable. Both live in the same workflow, both get the same logs and event history, and neither requires you to provision anything.

The piece that makes this practical is managed authentication. Connecting an account to Pipedream stores the token and refreshes it, and any code step in the workflow can then reach that account without you writing an OAuth dance or storing a refresh token yourself. That single feature is why GTM engineers who are perfectly capable of writing the integration themselves still use Pipedream: the interesting work is the enrichment logic or the routing rules, not the credential plumbing behind seven vendors.

Since 2024 the company has pushed hard in two directions. Connect turns the internal integration catalog outward, so a SaaS product can embed Pipedream's connectors and let its own users authorize Slack or HubSpot inside that product rather than building each integration in-house. Pipedream also runs hosted MCP servers exposing roughly 10,000 tools, which is how an AI agent gets safe, authenticated access to a customer's stack without a bespoke tool definition per API.

The important status fact as of August 2026: Pipedream was acquired by Workday. The deal was announced in November 2025 and closed in Workday's fiscal fourth quarter ending January 2026, and pipedream.com now carries a "Pipedream has joined Workday" notice. The self-serve platform, the free plan, and the published rate card are all still operating unchanged, and Workday has publicly leaned on the builder community as a reason for the purchase. Buyers should nonetheless price in the usual acquisition risk: the strategic center of gravity is now Workday's agentic stack for HR and finance customers, not the independent developer platform.

Best for

Technical GTM and RevOps engineers, and small product teams, who want the connector catalog and hosted auth of an iPaaS but refuse to express real logic as configured boxes. Also strong for SaaS companies that want to ship integrations to their own customers without building each one, via Connect.

Not the right fit for

Non-technical operators. The visual builder exists, but the product assumes you can read and write code, and the documentation is written for developers.
Interactive, latency-sensitive triggers such as Slack slash commands, where the cold start after roughly five minutes of inactivity adds seconds before your code runs.
Teams that want a predictable per-task bill. Credits meter compute, so a slow upstream API or a memory-heavy step costs more than a fast one doing the same job.
Long-running batch jobs. The ceiling is 750 seconds per execution, so anything longer has to be chunked into queued events.
Buyers who need a vendor with no ownership uncertainty; Pipedream is now a Workday property and the roadmap will follow Workday's priorities.
Marketing teams whose entire need is "when a form is submitted, add a row and send an email", which a simpler tool does with less setup.

How it works

1
You create a workflow and pick a trigger: an HTTP or webhook endpoint that Pipedream generates for you, a schedule, an inbound email address, or an app event such as a new HubSpot deal or a new row in Airtable. The webhook trigger is the one most GTM teams start with, because it turns any form, CRM automation, or third-party callback into an entry point in about a minute.
2
Steps run in order below the trigger. Each step is either a pre-built component from the public registry (a verified action such as "Send message to channel" for Slack) or a code cell you write. Data flows forward through a `steps` object, so step three reads whatever step two returned. Control flow is available as first-class blocks: filters, delays, if/else, switch, and parallel branches.
3
Authentication is per connected account, not per workflow. You link an app once, Pipedream keeps the token fresh, and both configured actions and your own code steps use it by reference. Environment variables and secrets cover anything not in the connector catalog.
4
Execution happens on managed workers. Memory is configurable up to 10 GB and duration up to 750 seconds on paid plans (300 on Free), with 2 GB of ephemeral file system per execution. Credits meter that compute: one credit is 30 seconds at the 256 MB default, so raising memory raises credit burn proportionally.
5
Every event is recorded. The event inspector shows the exact payload, the return value of each step, and the logs, which is the main reason debugging on Pipedream feels closer to a local dev loop than to a no-code canvas. History is retained 7 days on Free and Basic, 30 days on Advanced and Connect.
6
For production work you layer on the operational features: auto-retry on error, concurrency and execution rate controls, a queue with a configurable maximum, error notifications to email or Slack, and bi-directional GitHub sync so workflows live in a repository, get reviewed as pull requests, and can be edited locally.

Feature breakdown

25 features in 4 modules

Workflow builder

The core authoring surface: triggers, steps, and control flow, with code allowed anywhere.

Multiple triggers per workflow: One workflow can be fired by a webhook, a schedule, and an app event at once, so you are not maintaining three near-identical copies of the same logic.
Verified triggers and actions: A public registry of maintained components across 3,000+ apps, each a normal configured step with fields, so common operations need no code at all.
Code steps in four languages: Node.js, Python, Go, and Bash steps sit inline alongside configured actions, with most npm, PyPI, and Go packages importable by name without a build step.
Branching and control flow: Filters, if/else, switch, parallel branches, and delays are native blocks rather than tricks, which keeps non-trivial routing readable.
HTTP request builder: Build any API call against a connected account, with auth injected automatically, and scaffold the request into a reusable code step when you outgrow the form.
Data stores: A simple key-value store scoped to your workspace for deduplication, cursors, and small state, with limits rising from 50 keys on Free to 500 and larger storage on paid tiers.
File stores: Persist files between executions or upload them manually, which is what makes multi-step document and export workflows possible on an ephemeral runtime.

Triggers and event handling

How work enters the platform, and what happens when volume or failures arrive.

Generated HTTP endpoints: Every workflow can expose a unique URL that accepts any payload, with custom HTTP responses so the workflow can act as a small API rather than only a sink.
Schedule triggers: Cron-style scheduling with a floor of 5 minutes on Free, 1 minute on Basic and Advanced, and 1 second on Connect and Business.
Email triggers: Each workflow can receive a dedicated inbound email address, useful for parsing vendor notifications that offer no webhook.
Auto-retry on error: Failed executions retry automatically on paid plans, which is the difference between a workflow that silently drops events and one you can put in front of revenue data.
Concurrency and rate controls: Cap how many executions run in parallel and how fast the queue drains, so a burst of 5,000 webhooks does not get your workflow rate-limited out of a downstream API.
Event queue and inspector: A real-time inspector shows every event with full payloads and per-step return values; queue depth is capped at 100 events on Free and Basic and 10,000 on Advanced and above.

Connect and MCP

The outward-facing half: embed Pipedream's integrations in your own product or agent.

Managed authentication for your users: Your customers authorize Slack, Google, or HubSpot through Pipedream's flow, and your application gets a token reference instead of an OAuth implementation per vendor.
Custom OAuth clients: Ship the integration under your own OAuth app so the consent screen shows your brand rather than Pipedream's, available on paid tiers.
10,000 pre-built tools: The same component registry exposed programmatically, so a product or agent can call a named action against a connected account without writing the API client.
Hosted and self-deployed MCP servers: Deploy MCP servers over SSE or streamable HTTP that expose those tools to an LLM agent with authentication already handled per end user.
API proxy: Make authenticated requests to any connected app through Pipedream, which keeps customer tokens out of your own infrastructure.
Run user workflows: Trigger a Pipedream workflow on behalf of a specific end user, which is how embedded automations stay scoped to the right account.

Development workflow and operations

The parts that decide whether this survives contact with a real engineering process.

Bi-directional GitHub sync: Workflows live in a repository, sync both ways, and support branches, diffs, pull requests, and local editing; a workspace setting can require GitHub sync for all changes.
Environment variables and secrets: Workspace-scoped secrets available to every step, so credentials outside the connector catalog do not end up pasted into code.
Database connections: A built-in SQL editor, schema explorer, and prepared statements, plus a shared static IP so a locked-down database can allowlist Pipedream.
Open-source component registry: Triggers and actions are public on GitHub, so when a component is missing a field you can read the source, fork it, and publish a custom version.
Access controls and roles: Admin and editor roles, per-project and per-connected-account access controls, and enforced two-factor authentication on higher tiers.
Eliminate cold starts: Paid plans can keep workers warm, which removes the multi-second first-request penalty that makes Free-tier workflows feel sluggish on interactive triggers.

Use cases

4 documented

RevOps engineer at a 40-person B2B SaaS company

Inbound demo requests need company enrichment, ICP scoring against rules the CRM cannot express, routing to the right rep by territory and capacity, and a Slack alert, all before the lead goes stale.

One workflow triggered by the form webhook: an enrichment API call, a Python scoring step holding the actual rules, an if/else on the score, a HubSpot action to create and assign the deal, and a Slack action. The scoring logic is 30 lines of readable code instead of nine chained filter branches, and GitHub sync means it gets reviewed like the rest of the codebase.

Founder of a small SaaS product with an integrations backlog

Prospects keep asking for Slack, Google Sheets, and HubSpot integrations, and each one is a week of OAuth work plus ongoing token maintenance that a two-engineer team cannot afford.

Connect handles the authorization flow and token refresh, with a custom OAuth client so the consent screen carries the company's own branding. Three integrations ship in the time one would have taken, and the marginal cost of the fourth is close to zero.

Growth engineer building an AI research agent

An internal agent needs to read the CRM, search a knowledge base, and post to Slack, but hand-writing tool definitions and per-user auth for each API is the entire project.

The agent points at a Pipedream MCP server and gets authenticated tools across the stack immediately. Engineering time goes into prompt design and guardrails rather than into OAuth handling, and adding a fourth system is a configuration change.

Data-minded marketer replacing a brittle Zapier stack

A chain of eleven Zaps handles list hygiene and event syncing, the task bill climbs with volume, and one loop consumes tasks on records that are then discarded.

The chain collapses into two Pipedream workflows where the filtering happens inside a single code step. Because credits meter compute rather than steps executed, discarding 90 percent of records early costs almost nothing, and the monthly bill stops tracking record volume.

Pricing

from Free (100 credits per month, hard-capped); paid plans from $29 per month billed annually, or $45 month-to-month

Subscription plus usage. Every plan includes a monthly allowance of credits, where one credit is 30 seconds of execution at the default 256 MB of memory, and credit burn scales with the memory you configure. Plans also include AI token allowances for the AI-assisted builder. Tiers gate active workflow count, connected accounts, event history retention, schedule granularity, retries, GitHub sync, and security controls. Prices below are the annual rate; monthly billing costs meaningfully more.

Plan	Price	Includes
Free	$0 per month	100 credits per month, enforced as a hard cap rather than an overage 3 active workflows and 3 connected accounts Up to 300 seconds execution duration and 10 GB memory 7-day event history, 5-minute minimum schedule, community support Genuinely usable for learning and for one or two low-volume workflows, but the 3-workflow ceiling is what pushes most people to upgrade, not the credits.
Basic	$29 per month, billed annually ($348 per year; $45 month-to-month)	2,000 credits and 20M AI tokens per month 10 active workflows and 5 connected accounts 750-second execution duration, auto-retry on error, warm workers 1-minute minimum schedule interval, 7-day event history, email support
Advanced	$49 per month, billed annually ($588 per year; $74 month-to-month)	2,000 credits and 50M AI tokens per month Unlimited active workflows and connected accounts, plus workspace users Bi-directional GitHub sync with branches, diffs, and pull requests 30-day event history, 10,000-event queue, concurrency and rate controls The practical starting point for a team. Note that Advanced buys unlimited workflows and GitHub sync, not more credits: the included allowance is the same 2,000 as Basic.
Connect	$99 per month, billed annually ($1,188 per year; $150 month-to-month)	10,000 credits per month and 50M AI tokens Everything in Advanced plus the full Connect feature set for embedding integrations in your own product Custom OAuth clients, API proxy, deployed MCP servers, run user workflows 1-second minimum schedule interval Aimed at developers shipping integrations to their customers, not at internal automation. On Free and Basic, Connect is available in development mode only.
Business	Custom annual contract, quoted by sales	Custom credit and token volumes, custom event history retention HIPAA, SLA, SSO, SCIM provisioning, restricted login methods Virtual private clouds, unique egress IPs, VPC peering Dedicated Slack Connect channel, dedicated success engineer, custom invoicing and contracts

Billing notes

The meter is compute time, not tasks. A workflow that calls a slow API for 90 seconds costs three credits at default memory; a workflow that filters out a record in 200 milliseconds costs a fraction of one. This is the single biggest difference from per-task pricing, and it favors workloads that discard most events early.
Raising a step's memory raises credit burn proportionally. A step configured at 1 GB consumes credits roughly four times as fast as the same step at the 256 MB default, which is a common cause of unexplained overage.
The Free plan is hard-capped at 100 credits per month, so it stops rather than bills you. Paid plans do not stop: credits beyond the included allowance are charged as overage on the following invoice, so a runaway retry loop is a financial event, not just an error.
Advanced costs $20 per month more than Basic but includes the same 2,000 credits. You are paying for unlimited workflows, unlimited connected accounts, GitHub sync, and 30-day history.
Annual billing is a large discount, not a token one: Basic is $29 per month annually against $45 monthly, and Connect is $99 against $150.
Unused credits do not roll over between periods.
AI token allowances are for the AI-assisted builder and are separate from credits; additional tokens can be purchased.

Value assessment: For a technical buyer, Pipedream is one of the better value propositions in this category, because the pricing axis rewards writing efficient code instead of punishing you for having many steps. A team replacing a per-task automation bill that scales with record volume often sees the number fall sharply, since filtering happens inside one credit-cheap step rather than across a dozen billed actions. The caveats are real: the included credit allowance stops improving between Basic and Advanced, overages on paid plans are uncapped, and the cost of a workflow is a function of upstream API latency you do not control. Compared with self-hosting n8n, you are paying roughly $600 a year to not run infrastructure, which is a straightforward trade for most small teams and a bad one for teams that already have a Kubernetes cluster and a spare engineer.

Strengths & limitations

Strengths

Arbitrary Node.js, Python, Go, and Bash inline with configured actions, so no logic problem forces you off the platform.
Managed authentication across 3,000+ apps removes the single most tedious part of building integrations, and code steps get the same credentials as configured ones.
Compute-based pricing is unusually favorable for high-volume, high-filter workloads where most events are discarded early.
Bi-directional GitHub sync with branches and pull requests makes automation reviewable like the rest of the codebase, which very few tools in this category offer.
The event inspector shows full payloads and per-step return values, giving a debugging loop closer to local development than to a no-code canvas.
Connect and the hosted MCP servers turn the connector catalog outward, letting a small product ship customer-facing integrations or authenticated agent tools without building each one.
The component registry is open source on GitHub, so a missing field in a connector is a fork away rather than a support ticket.
A permanent free plan and transparent published prices through the $99 tier; nothing before Business requires talking to sales.

Limitations

Cold starts of several seconds after roughly five minutes of inactivity make Free-tier workflows unsuitable for interactive triggers such as Slack slash commands; warm workers are a paid-tier feature.
The 750-second execution ceiling (300 on Free) means genuinely long batch jobs must be chunked into queued events rather than run as one pass.
Advanced includes no more credits than Basic, which surprises buyers who upgrade expecting headroom and get feature unlocks instead.
Paid-plan overages are uncapped, so a retry loop against a slow API can burn thousands of credits overnight and appear on the next invoice.
Credit cost depends on memory configuration and upstream API latency, which makes forecasting the monthly bill harder than a per-task model.
Free and Basic tiers cap event history at 7 days, which is short if you investigate a data discrepancy a week and a half after it happened.
Without a custom OAuth client (paid tiers), end users authorizing through Connect see Pipedream's name on the consent screen rather than yours.
It is a developer tool with developer documentation; a non-technical operator will not be productive in it, whatever the visual builder suggests.
Ownership changed in early 2026. Workday has committed to supporting existing customers, but the roadmap now serves an enterprise agentic stack, and small self-serve buyers are not the strategic center of that business.

Head-to-head comparisons

6 alternatives

Pipedream vs Activepieces

from Free on cloud (daily credit allowance, one user) or free forever self-hosted; Plus from $16 per month billed yearly

Both let you drop code into a workflow, but they solve different problems. Activepieces is open source and self-hostable, which matters when data residency or an air-gapped environment is a requirement, and its pricing is per task on cloud. Pipedream is closed-source and hosted only, but has a far larger connector catalog, managed auth you can call from code, and Connect for embedding integrations in your own product. Choose Activepieces if self-hosting is the point; choose Pipedream if the connector breadth and the hosted runtime are.

Full Pipedream vs Activepieces comparison

Pipedream vs Relevance AI

from Free (200 Actions per month); Pro from $19 per month billed annually

Different layers of the same stack. Relevance AI builds AI agents and agent teams as the primary object, with tools and memory arranged around them. Pipedream builds deterministic event-driven workflows and exposes integrations to agents through MCP. If you want an agent that reasons about a task, Relevance AI is the shorter path; if you want a workflow that runs the same way 40,000 times and occasionally calls a model, Pipedream is. Teams often run both, with Pipedream supplying the authenticated tools.

Full Pipedream vs Relevance AI comparison

Pipedream vs n8n

from About $24/mo (Cloud Starter, 2,500 executions; roughly $20 on annual billing)

The closest comparison, and the decision usually comes down to hosting and pricing shape. n8n is source-available and can be self-hosted for the cost of a server, with a node-graph canvas that handles branching and merging more visually than Pipedream's linear-with-branches builder. Pipedream is hosted only, has more pre-built connectors, and bills compute rather than executions. Teams with infrastructure and a strong preference for owning their data pick n8n; teams that want to write code without running anything pick Pipedream.

Full Pipedream vs n8n comparison

Pipedream vs Zapier

from $19.99/mo (Professional, 750 tasks, annual billing)

Not really competitors for the same buyer. Zapier is built so a marketer can ship an automation with no engineer, and it charges per task. Pipedream assumes you write code and charges for compute. On identical work Pipedream is usually far cheaper at volume, especially when most events get filtered out early, but it demands technical ownership Zapier does not. A useful rule: if the person maintaining the automation cannot read a stack trace, buy Zapier.

Full Pipedream vs Zapier comparison

Pipedream vs Make

from $9/mo (Core, 10,000 operations, annual billing)

Make offers a much richer visual canvas with iterators, aggregators, and error handlers as drawn objects, and it is genuinely usable by a non-developer who is willing to learn it. Pipedream trades that visual expressiveness for the ability to write real code with real packages. Make's operations-based pricing rewards compact scenarios; Pipedream's credit pricing rewards fast ones. For a technical team, Pipedream's version control story is the decisive advantage Make does not match.

Full Pipedream vs Make comparison

Pipedream vs Gumloop

from $37 per month (Pro, 20,000 credits included)

Gumloop is built around AI-native workflows for operators, with LLM steps as the centerpiece and a canvas designed for people who are not engineers. Pipedream is general-purpose infrastructure that happens to support AI steps and MCP. If the workflow is mostly "read these documents, extract this, write it somewhere", Gumloop gets there faster. If the workflow has real API surface, authentication complexity, and needs to be reviewed in a pull request, Pipedream is the correct tool.

Full Pipedream vs Gumloop comparison

Implementation & onboarding

Setup time: A first working workflow takes 10 to 20 minutes: sign up, create a trigger, connect one app, add a step. A production-grade workflow with retries, concurrency limits, error alerting, and GitHub sync is a half day. Connect implementations are a genuine engineering project, typically one to two weeks for a first customer-facing integration.
Learning curve: Low for anyone who writes JavaScript or Python; the platform mostly gets out of the way. The parts that take longer are the ones specific to the model: understanding what a credit actually costs at your memory setting, deciding when to use a data store rather than an external database, and structuring long work as queued events instead of one long execution.
Onboarding: Entirely self-serve up to the Connect tier, with extensive developer documentation, a public component registry on GitHub, a changelog, and both a community forum and a public Slack. Business customers get a dedicated Slack Connect channel and a named success engineer.
Migration notes: There is no importer from Zapier, Make, or n8n; migration means rebuilding, though the rebuild is often shorter than expected because several chained steps collapse into one code step. Move the highest-volume workflows first, since those are where compute pricing pays off. Before cutting over, set concurrency and rate limits and confirm auto-retry behavior, because Pipedream will happily accept a burst that your downstream API will not. Export anything you need from event history first: it ages out at 7 or 30 days depending on tier.

Platform, API & security

Platforms: Web applicationREST APICLINode.js and Python SDKs for ConnectHosted and self-deployed MCP servers
API: A full REST API covers workflows, projects, connected accounts, and event data. Connect ships client and server SDKs for embedding managed authentication and running components on behalf of end users, plus an API proxy for authenticated requests to any connected app. The component registry is open source, so triggers and actions can be forked and published.
Compliance: SOC 2 Type IIGDPRHIPAA (Business plan)
Data residency: US-based infrastructure by default; virtual private clouds, VPC peering, unique egress IPs, and data retention controls are Business-plan features.
SSO: Single sign-on, restricted login methods, and SCIM user provisioning on the Business plan; enforced two-factor authentication is available on lower paid tiers.
Security notes: Credentials for connected accounts are stored and refreshed by Pipedream rather than by your code, and the API proxy exists specifically so end-user tokens in a Connect implementation never touch your own servers. Access controls scope connected accounts and projects to specific members. A shared static IP lets a firewalled database allowlist Pipedream workers without opening the database to the internet.

Support & resources

Channels: Community forumPublic SlackEmail support (paid plans)Dedicated Slack Connect channel and success engineer (Business)
Documentation: Thorough developer documentation covering the runtime, limits, credit accounting, Connect, and MCP, with a public changelog. It is written for engineers and does not pretend otherwise.
Community: An active public Slack and forum, plus a large open-source component repository on GitHub where users file and often fix connector issues themselves. Community-contributed components are a meaningful part of the 3,000+ app catalog.

Company

Founded: 2019
Headquarters: San Francisco, California
Ownership: Acquired by Workday (announced November 2025, closed early 2026)
Founders: Tod Sacerdoti, Jason Endo, Dan Hsiung, TJ Koblentz
Employees: Not disclosed since the Workday acquisition
Funding: Raised a $20M Series A led by True Ventures in May 2022, with CRV, Felicis, and World Innovation Lab participating, before being acquired by Workday.

Funding history

Round	Amount	Year	Notes
Series A	$20M	2022	Led by True Ventures, with CRV, Felicis Ventures, World Innovation Lab, and more than 20 software executives participating.

Timeline

2019Founded in San Francisco by Tod Sacerdoti and a group of senior product and engineering people from BrightRoll, which Yahoo had acquired in 2014.
2022Raises a $20M Series A led by True Ventures, positioning the platform as developer-first integration infrastructure rather than a no-code automation tool.
2024Pipedream Connect launches, exposing the connector catalog and managed authentication as an SDK so other products can embed integrations for their own users.
2025Hosted MCP servers ship in March, giving AI agents authenticated access to thousands of tools; String.com launches mid-year as a prompt-driven agent builder on top of the platform.
2025Workday signs a definitive agreement to acquire Pipedream in November, citing the connector catalog and builder community as the basis for its agentic integration strategy.
2026The acquisition closes in Workday's fiscal fourth quarter ending January. The self-serve platform, free plan, and published pricing continue unchanged through August 2026, with the site carrying a "Pipedream has joined Workday" notice.

Integrations

Slack
HubSpot
Salesforce
Google Sheets
Airtable
Notion
GitHub
Stripe
OpenAI
Anthropic
Twilio
Discord
Jira
Asana
Google Drive
AWS
PostgreSQL
Shopify

Frequently asked questions

13 questions

What is Pipedream used for?

Building event-driven workflows that connect APIs and services. A workflow starts from a webhook, schedule, email, or app event, then runs a sequence of steps that can be pre-built actions for any of 3,000+ apps or your own Node.js, Python, Go, or Bash code. Common uses are lead routing and enrichment, syncing data between a CRM and other systems, internal alerting, and giving AI agents authenticated access to a company's tools.

How much does Pipedream cost?

There is a permanent free plan with 100 credits a month, 3 workflows, and 3 connected accounts. Paid plans billed annually are Basic at $29 a month, Advanced at $49, and Connect at $99, which work out to $348, $588, and $1,188 a year. Month-to-month costs considerably more: $45, $74, and $150. The Business plan is quoted by sales.

What is a Pipedream credit?

A unit of compute, not a unit of work. One credit buys 30 seconds of execution at the default 256 MB of memory per workflow segment. If you raise a step to 1 GB it burns credits about four times as fast. This means a workflow's cost depends on how long it runs, which includes time spent waiting on someone else's slow API.

Is Pipedream free?

Yes, permanently, within limits: 100 credits a month, 3 active workflows, 3 connected accounts, 7-day event history, a 5-minute minimum schedule interval, and no auto-retry. The Free plan is hard-capped rather than billed for overage, so it stops rather than surprising you. For most people the 3-workflow ceiling is the binding constraint, not the credits.

Who owns Pipedream now?

Workday. The acquisition was announced in November 2025 and closed in Workday's fiscal fourth quarter ending January 2026, and pipedream.com carries a notice that Pipedream has joined Workday. The self-serve platform, free plan, and published rate card were all still running unchanged as of August 2026, but the strategic direction now follows Workday's agentic stack for HR and finance customers.

Pipedream vs Zapier: which is cheaper?

Pipedream, usually, once volume is real, because Zapier bills per task executed and Pipedream bills for compute time. A workflow that receives 50,000 events and discards 45,000 of them early costs Zapier a task per step per event and costs Pipedream a fraction of a credit each. The trade is technical ownership: Zapier is built for non-developers and Pipedream is not.

Pipedream vs n8n: which should I pick?

If you want to self-host and own your data, n8n, which is source-available and runs on your own server. If you want a hosted runtime with more pre-built connectors and no infrastructure to maintain, Pipedream. n8n's node canvas is more visual for complex branching and merging; Pipedream's advantage is writing real code with real packages and syncing workflows to GitHub with pull requests.

Can non-developers use Pipedream?

Partially. Verified triggers and actions are configured through forms, so simple workflows need no code. But the documentation, the error messages, and the credit model all assume an engineer, and the moment a workflow needs anything unusual you are in a code step. Teams that want a non-technical operator to own automations should look at Make or Zapier instead.

What is Pipedream Connect?

The embedded version of the platform. Instead of using the integrations yourself, you let your own customers authorize their accounts through Pipedream and then call actions on their behalf from your product, without implementing OAuth per vendor. It includes custom OAuth clients so the consent screen carries your brand, an API proxy so tokens never touch your servers, and deployed MCP servers. It is the $99-a-month tier and above; Free and Basic get development mode only.

Does Pipedream work with AI agents and MCP?

Yes, and this is now a major part of the product. Pipedream runs hosted MCP servers, and lets you deploy your own, exposing roughly 10,000 pre-built tools across 3,000+ apps to an LLM agent with per-user authentication already handled. In practice it means an agent gets safe access to a customer's Slack, CRM, and drive without you writing a tool definition and an OAuth flow for each one.

How long can a Pipedream workflow run?

Up to 300 seconds on the Free plan and up to 750 seconds on paid plans, with memory configurable to 10 GB and 2 GB of ephemeral file system per execution. Work that exceeds the ceiling has to be split into chunks that queue as separate events, which is a real design constraint on large batch jobs.

What happens if I go over my credit allowance?

On the Free plan nothing is charged; you hit a hard cap and executions stop until the period resets. On paid plans there is no cap: credits beyond your allowance are billed as overage on the following invoice. That asymmetry matters, because a workflow retrying against a slow or failing API can consume a large number of credits overnight. Set concurrency and rate controls before putting anything high-volume into production.

Does Pipedream support version control?

Yes, on the Advanced plan and above, through bi-directional GitHub sync. Workflows live in a repository, you can develop in branches, view diffs, open pull requests, merge from GitHub, and edit locally, and a workspace setting can require GitHub sync for all changes. This is the clearest differentiator against Zapier and Make for an engineering team.

Editorial verdict

Pipedream is the automation platform for people who can write code and are tired of pretending they cannot. The combination of a 3,000-app connector catalog, managed authentication that your own code can use, a hosted runtime with no infrastructure to operate, and GitHub sync with pull requests is not matched by anything else at this price. Compute-based pricing is the right shape for GTM workloads that ingest a lot and keep a little, and the published tiers stay honest all the way to $99 a month before sales gets involved. The reservations are specific rather than generic: cold starts rule out latency-sensitive triggers on lower tiers, the 750-second ceiling forces awkward chunking on batch work, Advanced buys features rather than credits, and uncapped paid-plan overages punish a retry loop. Weigh those against the ownership change, since Workday closed its acquisition in early 2026 and an independent developer platform is now a component of an enterprise agentic strategy. For a technical small team today it remains an excellent buy; keep an eye on the roadmap and the renewal terms.

Written by the SaaSTracker editorial team. Awards, when shown, are judged against the published criteria in our methodology.