Pipedream

What Pipedream actually is

Pipedream is a developer-first integration and workflow automation platform. Think of it as serverless glue code without the serverless ceremony: you build event-driven workflows — triggered by webhooks, schedules, app events, or HTTP requests — out of pre-built actions from thousands of connected apps, or your own code in Node.js, Python, Go, or Bash. The platform runs those workflows in isolated virtual machines on AWS, manages OAuth for the connected services, and handles queues, retries, and logging.

The company was founded in 2019 in San Francisco by Dylan Sather and Tod Sacerdoti, raised a $20M Series A in 2021/2022 (True Ventures, with Y Combinator and Workday Ventures among the backers), and was acquired by Workday — agreement announced November 2025, closed February 2026. As of late 2026 the product still operates standalone with its existing pricing and community, but an acquisition is a fact we would weigh when betting a production integration pipeline on any vendor. We come back to that under trade-offs.

What distinguishes Pipedream from no-code automation tools is the code footing. Every step can be a real code block with npm and PyPI dependencies, secrets, environment variables, file system access (2 GB), and up to 10 GB of RAM per execution — not a formula editor with delusions of grandeur. Conversely, what distinguishes it from writing your own integration service is everything around the code: 2,500+ integrated apps with managed authentication (the Connect product advertises 3,000+ connectors and 10,000+ prebuilt tools and triggers), event sources for popular APIs, a component model you can publish back to the community, data stores, file stores, and more recently MCP servers so AI agents can call the same tool library.

The product surface now spans three things: Workflows (the original automation engine), Connect (an SDK and API for embedding Pipedream's integrations and managed auth into your own app or AI agent — this is where the growth push is), and agent tooling (MCP servers, an AI agent builder). If you evaluate Pipedream today, you are evaluating an integration platform that increasingly sells itself as "the integration layer for AI agents", not just a Zapier-for-developers.

Pricing: what you really pay

Pricing is credit-based, and the credits are defined in compute terms rather than task counts: one credit per 30 seconds of compute at 256 MB memory per workflow segment. Doubling memory doubles credit burn; steps are free, and development/testing executions cost nothing. That model is more honest than per-task pricing — a five-step workflow that runs for two seconds bills the same as a one-step one.

The Workflows plans: Free at $0 with no credit card, up to 300 credits/month (capped at 10 credits/day), 3 active workflows, 3 connected accounts, 1 data store, and access to the full 2,500+ app library and 10k+ triggers and actions. Basic at $29/month billed annually ($348) includes 2,000 credits/month, 10 active workflows, 5 connected accounts, and 1-minute schedule intervals. Advanced at $49/month billed annually ($588) also includes 2,000 credits but removes the structural limits: unlimited workflows and connected accounts, logical branching, concurrency and execution-rate controls, auto-retry, a shared static egress IP, custom OAuth clients, unlimited data stores, file stores, and 1-second schedule intervals. Business is contact-sales.

Pipedream Connect is priced separately: free in development, then Startup at $99/month billed annually ($150 monthly) with 100 external users included (then $2 per user), 10,000 usage credits/month (then $0.012 per credit), and email support; Business is custom with volume pricing, SSO, SCIM, HIPAA, and SLAs.

Our read: the free plan is genuinely useful — this is one of the few developer tools where you can run real, always-on automation without paying — and the $29 Basic tier is enough for side projects. The awkward middle is Advanced: $49/month buys you 2,000 credits and the features that make workflows production-grade (branching, retries, rate controls). If your automation is chatty, credits vanish faster than you expect, and there is no overage story below Advanced/Connect. Budget controls exist on Basic and up. Compared with running the same glue on Lambda plus your own scheduler, Pipedream costs more per compute-second and far less per engineering-hour.

Performance & infrastructure

There is no hardware to benchmark in the usual sense — see /benchmarks for our VM numbers instead — but the execution environment is well specified. Workflows run in isolated VMs on AWS (not classic function containers), which is why Pipedream can offer things Lambda makes awkward: no cold starts, a writable file system, up to 300 seconds execution on Free (750 on paid plans), up to 10 GB RAM, npm/PyPI package installs, and long-running steps. Execution duration limits and memory multipliers are documented plainly, and credits scale with memory accordingly.

Transparency is a strength. The docs page spells out exactly what a credit is, that development and test executions are free, that idle workflows bill nothing, and that downgrading a workspace disables workflows using plan-gated features. SOC 2 Type 2, GDPR, and HIPAA (Business) are on the compliance sheet. Static egress IPs on Advanced and above matter more than people expect when the workflow calls an API with IP allowlisting.

What is not published: where workflows execute geographically, any uptime percentage for the workflow service, or a latency SLA. Pipedream's status page history is the practical reliability record. For most automation workloads that is acceptable; for EU data-residency requirements it is a real question to raise with them before you sign, particularly post-Workday.

The component ecosystem deserves credit in this section because it functions as the platform's real "infrastructure". 2,500+ apps with managed OAuth, thousands of verified triggers and actions, community-published components, GitHub sync with branch-based development and diffs, and a tool registry that doubles as an MCP tool source. When a platform maintains the boring parts of 2,500 API integrations for you, that is the product.

Where it falls short

The acquisition is the strategic risk. Workday bought Pipedream for agentic integration technology; standalone pricing is intact today, but enterprise owners eventually reshape product direction, packaging, and support tiers. We would keep our workflow definitions in GitHub (the sync exists for exactly this) so exit costs stay low.

Pricing cliffs: the free plan's 10-credits-per-day cap makes it a hobby tier in practice; the jump from Basic to Advanced is where production features start; and Connect's per-user pricing can get expensive at scale. Credit burn from memory-heavy or long-running steps is easy to underestimate — a workflow that fans out over hundreds of records is doing hundreds of credit-bills worth of compute.

Compared with a raw serverless platform, Pipedream is opinionated and proprietary. Workflow definitions, component APIs, and data stores are Pipedream constructs; you cannot lift them onto Lambda without rewriting. That is the trade for the convenience, and it is the same trade Zapier users make — just at a more technical level.

Finally, support tiers: community Slack and forum on Free, email on paid, dedicated Slack and SLAs only on enterprise contracts. For a product that runs mission-critical integrations, that middle tier is thin.

Who should pick this

We would pick Pipedream for developer teams drowning in integration plumbing: webhook receivers, sync jobs between SaaS products, event fan-out, internal tooling, and increasingly the auth and tool layer behind AI agents. It is our pick when the alternative is building and babysitting a queue, a scheduler, and 30 OAuth flows. The free tier is enough to prove out the pattern, and the code-in-steps model means you never hit the wall where the no-code tool cannot express your logic.

We would not pick it for high-volume batch compute (credits will punish you), for strict EU data residency without written confirmation of execution regions, or for organizations that want to own every line of the integration stack. If your workflows are simple app-to-app chains and nobody on the team writes code, a no-code tool will be cheaper to operate.

Key takeaways

• Code-first workflow automation: Node.js, Python, Go, or Bash steps over 2,500+ managed app integrations, with Connect now pushing 3,000+ connectors for embedded use.
• Credit pricing is compute-based (1 credit per 30s at 256 MB) — transparent, but memory-heavy and fan-out workflows burn credits fast.
• Free plan (300 credits/month, 3 active workflows, no credit card) is one of the few permanent free tiers in this category; Basic is $29/month, Advanced $49/month, billed annually.
• Execution environment is above function-class: no cold starts, 2 GB file system, up to 10 GB RAM, 750s runs, npm/PyPI packages.
• Owned by Workday since February 2026 — product and pricing intact today, but keep workflows in GitHub sync for portability.
• No published uptime SLA or execution-region disclosure outside enterprise contracts; check EU data-residency needs before committing.

Sources

• https://pipedream.com/pricing
• https://pipedream.com/pricing?plan=Connect
• https://pipedream.com/docs/pricing
• https://docs.pipedream.com/pricing
• https://linkedin.com/company/pipedreamhq
• https://topickz.com/software/pipedream-com
• https://akta.pro/company-profiles/0000akn/pipedream
• https://indexed.vc/companies/pipedream
• https://recatools.com/ai-directory/pipedream
• https://costbench.com/software/ai-automation/pipedream