Back to vendors
P

Pipedream

Also known as: Pipedream Connect, Pipedream MCP, MCP Chat

Visit site
Entry priceFree tier reported at around 100 credits per day; Basic around $29 per month; Advanced around $79 per month; Business and enterprise customFull pricing detail

Developer first integration platform whose 3,000 app catalogue doubles as agent tooling, exposing 10,000 prebuilt tools through a mature MCP server with managed OAuth, plus an embeddable toolkit for putting integrations inside your own agent.

Pipedream is a developer first integration and workflow platform from San Francisco that has become one of the most consequential pieces of agent tooling infrastructure. It pairs a visual workflow builder with inline Node.js, Python, Go and Bash, so dropping into code is a first class option rather than an escape hatch, and it bills per compute second rather than per task.

What makes it matter for agents is that its integration catalogue doubles as agent tooling. The same 3,000 plus APIs and 10,000 plus prebuilt components that power workflows are exposed as Model Context Protocol tools, with managed OAuth, per user credentials and encrypted credential storage handled by the platform. Its MCP server is generally available and is among the most mature production MCP deployments in the market, runnable as a remote service or self hosted through npx or Docker, and the company publishes MCP Chat as an open source reference agent implementation.

Pipedream Connect is the second half of the story: an embeddable developer toolkit for putting integrations inside someone else's product or agent. It covers managed authentication with approved OAuth client IDs, durable components, prebuilt triggers and actions exposed as agent tools, a Connect Proxy for arbitrary API calls without handling auth, and hosted MCP servers. Alongside it, an Agent Builder lets teams prompt, run, edit and deploy agents with code generation for any API, backed by built in queues, data stores and one click private networks. Agents generally run as supervised automations rather than fully unsupervised actors.

The corporate fact to carry: Workday signed a definitive agreement to acquire Pipedream on 19 November 2025, expected to close by the end of January 2026, citing more than 5,000 customers and tens of thousands of users. The product has continued business as usual, but Workday's direction will shape the roadmap, which is worth weighing on any long horizon dependency.

Vendor details

Canonical URL

https://pipedream.com

Category

Agent infrastructure

Company status

acquired

Use cases & customers

Target customers

developersproduct teams embedding integrationsAI agent builders

Deployment options

SaaScloudself-hosted MCP serverprivate networks

In practice

Your agent needs to act across Slack, GitHub, Stripe and a few hundred other systems and you do not want to build or maintain OAuth for any of them. One MCP server endpoint exposes the whole catalogue with managed credentials.

You are shipping integrations inside your own product. Pipedream Connect provides the embeddable toolkit, approved OAuth client IDs and durable components so the integration surface is not a two year internal project.

A workflow needs logic no prebuilt action covers. Node.js, Python, Go or Bash run inline as a normal step, and billing is per compute second rather than per task.

Agentic Index coverage score

9.0 / 14 capabilities · 64%

Integrations & Tool CallingThe defining capability: more than 3,000 APIs with over 10,000 prebuilt tools, with managed OAuth, per user credentials and encrypted credential storage handled by the platform, plus a Connect Proxy for arbitrary API calls. Pipedream homepage and Connect page 2026-08-05 Full
Workflow OrchestrationA visual workflow builder where every step can be a prebuilt component or inline Node.js, Python, Go or Bash, backed by built in queues and durable components, plus an Agent Builder for prompting, running, editing and deploying agents. 2026-08-05 Full
Knowledge Grounding & RAGNo retrieval, document ingestion, vector store or knowledge grounding capability is documented. Data stores are key value state primitives rather than a grounding layer, and grounding is left to whatever agent framework sits on top. 2026-08-05 Unable to verify
Human Oversight & GuardrailsAgents are characterised as running as supervised automations rather than fully unsupervised actors, and workflows are deployed deliberately, but no approval gates, review queues or configurable guardrail policy is documented. 2026-08-05 Partial
Security, Identity & GovernanceManaged authorisation flows with approved OAuth client IDs, secure token storage and refresh, per user credentials, encrypted credential storage, and single sign on with SAML plus audit logging on the business tier. Credential handling is the product's core security surface, not an add on. 2026-08-05 Full
Observability & AuditabilityAudit logging is available on the business tier and per step workflow execution is inspectable during development, but no agent level tracing, run history analytics or evaluation reporting is documented as a first class surface. 2026-08-05 Partial
Memory & State PersistenceBuilt in data stores are a first class primitive for persisting agent state across runs, alongside durable components, built in queues and secure persistent credential storage with automatic token refresh. 2026-08-05 Full
Deployment & Data ResidencyThe MCP server can be self hosted via npx or a Docker container and one click private networks are offered, but the platform itself is managed cloud only, self hosting still requires Pipedream OAuth credentials, and no regional data residency option is documented. 2026-08-05 Partial
Prebuilt Agents, Templates & PacksMore than 10,000 prebuilt triggers and actions across 3,000 apps, publishable custom tools scoped to a workspace, and MCP Chat shipped as an open source reference agent implementation. 2026-08-05 Full
Triggers & Channel CoveragePrebuilt triggers across the full app catalogue alongside HTTP, webhook and schedule based invocation, and the same triggers can be exposed as tools to an agent rather than only starting a workflow. 2026-08-05 Full
Model Flexibility & RoutingThe Agent Builder and the edit and debug assistant use models, but no model list, routing policy, model selection or bring your own key capability is documented; model choice is left to whatever framework consumes the MCP tools. 2026-08-05 Unable to verify
APIs, SDKs & MCP ExtensibilityAmong the strongest extensibility surfaces in the index: an embeddable SDK in Pipedream Connect, remote and self hosted MCP servers exposing the whole catalogue, a Connect Proxy, publishable custom tools, and an open source reference agent. Extensibility is the product rather than a feature of it. 2026-08-05 Full
Testing, Debugging & OptimizationAn edit and debug with AI assistant helps build and repair workflows from natural language and steps can be run individually during development, but no evaluation harness, benchmark set or regression testing for agent behaviour is documented. 2026-08-05 Partial
Browser & Computer UseNo browser control or computer use capability. The platform reaches systems through authenticated APIs rather than by operating interfaces, which is the explicit architectural alternative to computer use. 2026-08-05 Unable to verify

The Agentic Index coverage score grades every vendor Full, Partial or Unable to verify against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded

Pricing

Free tier reported at around 100 credits per day; Basic around $29 per month; Advanced around $79 per month; Business and enterprise custom

credits, metered per compute second

Free tierTrial available

What is public

Tier structure and approximate rates circulate widely in third party analyses; the vendor publishes a pricing page that was not retrieved in this pass.

Billing mechanics

Credit based tiers where consumption is metered per compute second rather than per task or per operation. That is a meaningful structural difference from Zapier and Make style per task pricing: short, fast workflows cost materially less on this model, while long running or heavy compute steps cost more.

Cost watchouts

Single sign on, SAML, audit logging and unlimited credits sit behind the custom business tier, so governance requirements move a buyer straight off the published rates

Variable cost rationale

Compute second metering ties cost directly to workload shape rather than headcount, which is cheaper for fast workflows and more expensive for long running agent loops, and agent workloads are exactly the case where run duration is hardest to predict in advance.

Additional watchouts

The published rates were not confirmed first hand in this pass. Separately, the acquisition means roadmap direction now sits with Workday, which matters most for teams embedding Pipedream Connect into their own product rather than using it internally.

Overage / add-ons

Credit consumption scales with compute time; the business tier is described as offering unlimited credits

Sales call required

Mixed (some tiers require a call)

Free / trial

Free tier reported at approximately 100 credits per day as of 2026, sufficient to build and run real workflows before paying

Lowest paid plan

Basic

Commercial notes

The per compute second model is the commercial expression of the product's developer first stance, and it is the reason integration analyses recommend it for teams that can write code and run tight workflows. Buyers should also weigh the Workday acquisition, signed 19 November 2025 and expected to close by the end of January 2026, when making a long horizon dependency decision.

Key ambiguities

Every figure here comes from third party platform analyses rather than a retrieved first party pricing page, and the rates are described as approximate. Credit definitions and the compute second conversion are not confirmed. Re verify against pipedream.com pricing before publishing any number.

Missing data

Confirmed first party rates, the credit to compute second conversion, business tier pricing, and any post acquisition pricing changes under Workday.

Agentic Index verified 2026-08-05

Alternatives to Pipedream

The closest documented capability profiles to Pipedream among agent infrastructure platforms tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.

  • Temporal9.5 / 14Fuller documented coverage on Observability & Auditability and Deployment & Data Residency
  • Hatchet9.0 / 14Fuller documented coverage on Observability & Auditability and Deployment & Data Residency
  • Restate8.5 / 14Fuller documented coverage on Human Oversight & Guardrails and Deployment & Data Residency
  • Arcade8.0 / 14Adds documented Model Flexibility & Routing
  • DBOS9.0 / 14Adds documented Model Flexibility & Routing
  • Inworld AI Agent Runtime11.0 / 14Adds documented Knowledge Grounding & RAG and Model Flexibility & Routing

Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded

Contact us

Found a vendor we missed? Have feedback on the index? We'd love to hear from you.