Skip to content
Talk to Us

Blocks.ai publishes this comparison and sells in this market. Every claim about another product links to a public source.

DIY vs n8n: an in-house build or a platform for AI agents and workflows

Build it yourself (DIY)

Building agent connections and controls in-house from open protocols, existing infrastructure, and open-source tools

n8n

Platform for AI agents and workflows you can see and control, built visually or with code

Short answer

DIY is not a product: you build agents and their controls yourself, from open-source frameworks and open protocols such as MCP and A2A, which leave authorization logic to the implementer.⁠Source 1, Source 2, Source 3, Source 4 n8n says it is a product for building AI agents and workflows, visually or with code, managed or self-hosted; its Agents feature is in preview.⁠Source 5, Source 6, Source 7

By Blocks.ai teamLast checked

Where each one sits

Six layers of running AI agents at a company, and what each product’s own public sources say it covers.

These aren’t the same kind of product

Where they overlap
Both are ways to build AI agents: DIY with open-source frameworks such as LangGraph or Google’s ADK, n8n, it says, visually or with code.⁠Source 1, Source 5, Source 8
Where they differ
DIY also covers the controls around agents, such as Uber’s MCP registry and gateway.⁠Source 9 n8n’s agents, in preview, live in n8n projects; governing agents built elsewhere is not publicly documented.⁠Source 7
Running both
n8n’s docs say it can act as an MCP client and an MCP server, so it can work with MCP servers and agents a team built itself.⁠Source 10, Source 11
Public sources · checked 2 October 2026
  • Offered
  • Preview
  • You build it
  • Not publicly documented

DIY

  • Build agents: You build itOpen-source frameworks like ADK⁠Source 1
  • Host and run agents: You build itSelf-hosted, like LangGraph⁠Source 8
  • Identity and access: You build itYour own identity provider⁠Source 12
  • Registry and governance: You build itYour own agent registry⁠Source 13
  • Traffic between agents, tools, and models: You build itYour gateway and service mesh⁠Source 9
  • Agents across organizations: You build itA2A with API management⁠Source 14

n8n

  • Build agents: OfferedLangChain-based AI Agent node⁠Source 15
  • Host and run agents: OfferedYour infrastructure or n8n’s⁠Source 5
  • Identity and access: Not publicly documented
  • Registry and governance: PreviewAgents stored in projects⁠Source 7
  • Traffic between agents, tools, and models: OfferedMCP client for external tools⁠Source 10
  • Agents across organizations: Not publicly documented

At a glance

TopicDIYn8n
What it isAn in-house build from open protocols such as A2A and MCP, which the A2A specification calls complementary, and open-source agent frameworks.⁠Source 1, Source 2n8n says it builds AI agents and workflows, visually or with code.⁠Source 5
Where it runsWherever you run it: Pinterest optimized for MCP servers hosted in its internal cloud.⁠Source 16Fully managed n8n Cloud, hosted on Microsoft Azure, currently in the EU, or self-hosted on your own infrastructure.⁠Source 6, Source 17
Agent registryYours to build: A2A’s current specification prescribes no standard API for curated registries, and the official MCP Registry is in preview and does not support private servers.⁠Source 13, Source 18Agents, in preview, live in n8n projects.⁠Source 7 A registry of agents built outside n8n is not publicly documented.
Access controlA2A calls authorization logic implementation-specific.⁠Source 2 MCP makes authorization optional.⁠Source 19Project roles on all Cloud plans and self-hosted Registered Community, Business, and Enterprise.⁠Source 20 SAML sign-in on Business and Enterprise; OIDC and custom roles on Enterprise.⁠Source 20, Source 21, Source 22
PricingThe A2A and MCP specifications are openly licensed, A2A under Apache 2.0.⁠Source 2, Source 3 Build and running costs are not publicly documented.Paid plans are priced by monthly workflow executions, with unlimited users.⁠Source 23 Cloud Starter is $20 a month, billed annually; Enterprise is through sales; self-hosted Community is free.⁠Source 23, Source 24

What each one is

Build it yourself (DIY)

DIY means doing it without buying an agent platform: agents built on open-source frameworks, MCP and A2A wired together in-house, existing infrastructure stretched to cover agents, an in-house platform, or no central approach; before its gateway, Uber says, teams built integrations independently and duplicated infrastructure.⁠Source 9

n8n

n8n says it is a product for building AI agents and workflows, visually or with code, on n8n Cloud or self-hosted.⁠Source 5, Source 6 Its AI nodes implement LangChain’s JavaScript framework.⁠Source 15 Its Agents feature, in preview, keeps agents in n8n projects alongside workflows.⁠Source 7

The differences that matter

  1. Agent registry

    DIY

    You build it: A2A’s current specification prescribes no standard API for curated registries, and Uber and Pinterest each run their own MCP registry.⁠Source 9, Source 13, Source 16

    n8n

    Agents built in n8n, in preview, live in n8n projects.⁠Source 7 Registering or governing agents built elsewhere is not publicly documented.

    The official MCP Registry is in preview and does not support private servers; its docs recommend hosting your own private registry for those.⁠Source 18

  2. Access control

    DIY

    In A2A, each server authorizes under its own policies; MCP authorization is optional.⁠Source 2, Source 19 Pinterest governs almost every MCP call with end-user JWTs and mesh identities.⁠Source 16

    n8n

    Users sign in with SAML (Business and Enterprise) or OIDC (Enterprise) and get project roles.⁠Source 20, Source 21, Source 22 n8n says agent access, in preview, follows its roles.⁠Source 25

    Per-agent identity from n8n is not publicly documented; in preview, n8n says each agent tool runs with the credential attached to it.⁠Source 25

  3. Running and support

    DIY

    Support depends on each component: MCP SDKs are tiered partly by maintenance commitments, and the official MCP Registry, in preview, gives no uptime guarantees.⁠Source 26, Source 27

    n8n

    n8n Cloud is fully managed.⁠Source 6 Dedicated support, with guaranteed response times available through SLAs, comes only with Enterprise; Business is self-serve.⁠Source 23

For security teams

What a security review asks, answered from each vendor’s public documentation.

TopicDIYn8n
Network exposureMCP servers on Streamable HTTP expose an HTTP endpoint; A2A agents on HTTP use HTTPS URLs in production.⁠Source 2, Source 28The MCP Server Trigger exposes a URL.⁠Source 11 Cloud clients of the built-in MCP server need a public instance, n8n says.⁠Source 29
IdentityIn A2A, identity is established at the HTTP layer; credentials come out of band.⁠Source 2, Source 14 MCP authorization is optional.⁠Source 19SAML on Business and Enterprise, OIDC on Enterprise.⁠Source 21, Source 22 Per-agent identity from n8n is not publicly documented.
Access changes and revocationEach A2A server authorizes requests under its own policies; the specification calls that logic implementation-specific.⁠Source 2Revoking an OAuth MCP client disconnects it at once.⁠Source 29 Unpublishing an agent (preview) stops users, channels, and schedules running it.⁠Source 7
Audit trailA2A docs advise auditing task creation, critical state changes, and agent actions, and tracing, for example with OpenTelemetry.⁠Source 14On Enterprise, log streaming to syslog, a webhook, or Sentry, including three audit events for the instance MCP server.⁠Source 30
ComplianceSits with the implementer: A2A docs say to ensure compliance with regulations such as GDPR, CCPA, and HIPAA.⁠Source 14n8n says the company holds a SOC 2 Type 2 report, shared with enterprise customers; others get SOC 3.⁠Source 17, Source 31

Full comparison

18 criteria in five groups. Every cell links to its source, or says no public source answers it.

DIY and n8n compared on 18 criteria
DIYn8n
What it is
What it is and who it’s forAn in-house build on open protocols: A2A, an open standard for communication between agent systems, and MCP, which the A2A specification calls complementary.⁠Source 2n8n says it builds AI agents and workflows, visually or with code.⁠Source 5 It runs as fully managed n8n Cloud or self-hosted on your own infrastructure.⁠Source 6
MaturityVaries by component: MCP’s latest revision is 2026-07-28.⁠Source 4 The official MCP Registry is in preview and may have breaking changes or data resets.⁠Source 18n8n’s changelog dates Agents (preview) to September 2026; they are on every plan.⁠Source 7, Source 32 n8n made its LangChain-based AI nodes generally available in November 2023.⁠Source 33
Control
Agent registry and discoveryYours to build: A2A’s current specification prescribes no standard API for curated registries.⁠Source 13 Uber and Pinterest each describe their own MCP registry.⁠Source 9, Source 16Agents, in preview, live in an n8n project alongside workflows.⁠Source 7 A registry of agents built outside n8n is not publicly documented.
Identity and access controlIn A2A, identity is established at the HTTP layer and authorization logic is implementation-specific.⁠Source 2, Source 14 MCP authorization is optional.⁠Source 19Project roles set what a user can do in a project; self-hosted Community must register.⁠Source 20 SAML on Business and Enterprise, OIDC on Enterprise.⁠Source 21, Source 22
Ownership, policy, and revocationMCP says implementers should build consent and authorization flows.⁠Source 4 Uber starts every MCP server and tool disabled until its owning team reviews and enables it.⁠Source 9Human approval can be required before an AI Agent node runs a tool.⁠Source 34 Agents (preview) can be unpublished, so users, channels, and schedules can’t run them.⁠Source 7
Audit log and observabilityA2A docs advise auditing task creation, critical state changes, and agent actions.⁠Source 14 Pinterest’s MCP servers share library functions that log inputs and outputs.⁠Source 16On Enterprise, log streaming sends events to syslog, a webhook, or Sentry.⁠Source 30 Agent sessions (preview) show messages, tools used, and pending approvals.⁠Source 7
Connection
How agents connectMCP servers on Streamable HTTP expose an HTTP endpoint; A2A agents on HTTP use HTTPS URLs in production.⁠Source 2, Source 28 Uber’s gateway uses Uber’s service mesh sidecar.⁠Source 9The MCP Server Trigger node exposes a URL.⁠Source 11 For cloud-based clients of its built-in MCP server, n8n says to make the instance publicly accessible.⁠Source 29
Agents across organizationsA2A is designed for agents built by different companies on separate servers.⁠Source 35 Each server authorizes requests under its own policies.⁠Source 2The MCP Server Trigger node can expose n8n tools to external AI agents.⁠Source 10 A cross-organization access model is not publicly documented.
Protocol supportMCP defines stdio and Streamable HTTP transports.⁠Source 36 A2A maps its operations to JSON-RPC, gRPC, and HTTP/REST bindings.⁠Source 2Can act as an MCP client, and as an MCP server through a trigger node (SSE or streamable HTTP, not stdio).⁠Source 10, Source 11 First-party A2A support is not publicly documented.
Frameworks, models, and clouds supportedA2A gives agents built with different frameworks, languages, or vendors a common language.⁠Source 2 Google’s ADK says it is model-agnostic and deployment-agnostic.⁠Source 1AI nodes implement LangChain’s JavaScript framework, with your choice of agent, LLM, and memory.⁠Source 15 Managing agents built elsewhere is not publicly documented.
Operations
Deployment options and data residencyWherever you run it: Pinterest optimized for MCP servers hosted in its internal cloud, where its routing and security logic can be applied.⁠Source 16n8n Cloud runs on Microsoft Azure, with data stored in Frankfurt, Germany.⁠Source 17, Source 23 Self-hosted runs on your infrastructure, on-premises, or in a private cloud.⁠Source 37
Compliance attestationsSits with the implementer: A2A docs cite regulations such as GDPR, CCPA, and HIPAA; MCP leaves access controls and data protection to implementers.⁠Source 4, Source 14n8n says the company holds a SOC 2 Type 2 report.⁠Source 31 It is for enterprise customers; others can refer to the SOC 3 report.⁠Source 17
Support and SLADepends on the component: MCP SDKs are tiered partly by maintenance commitments.⁠Source 26 The official MCP Registry, in preview, gives no uptime guarantees.⁠Source 27Dedicated support, with guaranteed response times available through SLAs, on Enterprise only.⁠Source 23 Business is self-serve.⁠Source 23 Forum support through the n8n community.⁠Source 23
Time and effort to get runningA curated A2A registry is a service you deploy and maintain.⁠Source 13 Pinterest built a unified deployment pipeline after new MCP servers took too much setup.⁠Source 16A 14-day Cloud trial, or a one-step Docker setup.⁠Source 37, Source 38 Paid self-hosted plans use a license key that pings n8n daily, or a certificate when air-gapped.⁠Source 23, Source 24, Source 39
Pricing model and public pricesThe A2A and MCP specifications are openly licensed, A2A under Apache 2.0.⁠Source 2, Source 3 Build and running costs are not publicly documented.By monthly workflow executions, unlimited users.⁠Source 23 Billed annually: Cloud Starter $20 and Pro $50 a month.⁠Source 23 Self-hosted Community edition is free.⁠Source 24
Building
Agent building toolsOpen-source frameworks such as LangGraph and Google’s Agent Development Kit are for building and deploying agents.⁠Source 1, Source 8 The A2A project hosts SDKs in six languages.⁠Source 40n8n says you can use JavaScript or Python anywhere in a workflow.⁠Source 5 In Agent Builder (preview), agents get a model, instructions, tools, skills, and sub-agents.⁠Source 7
Model accessDepends on the framework: Google’s ADK, for example, says it is optimized for Gemini and model-agnostic.⁠Source 1AI workflows can connect providers such as OpenAI, Anthropic, and Google, and combine models.⁠Source 41 Bring-your-own-LLM uses your own API credentials.⁠Source 42
Integrations and ecosystemThe official MCP Registry, in preview, has a REST API for MCP clients and aggregators to discover servers.⁠Source 18Over 500 integrations, per n8n’s homepage, and 600+ agentic workflow templates, n8n says.⁠Source 5, Source 43 A built-in registry of MCP servers.⁠Source 44

Which to choose

Choose DIY if

  • You want your own control plane for agent tools, as Uber built with an MCP registry and a proxy gateway.⁠Source 9
  • You run a service mesh or proxy and want it to check MCP calls, as Pinterest does with Envoy and mesh identities.⁠Source 16
  • You need agents at other companies to reach yours: A2A is designed for agents from different companies on separate servers.⁠Source 35
  • You want your own review rules, as at Uber, where every MCP server and tool starts disabled until reviewed.⁠Source 9

Choose n8n if

  • You want to build AI agents and workflows visually or with code, with over 500 integrations, per n8n’s homepage.⁠Source 5
  • You’d rather not run it yourself: n8n Cloud is fully managed, with data stored in the EU, in Frankfurt.⁠Source 6, Source 23
  • You want single sign-on and roles built in: SAML on Business and Enterprise, OIDC and custom roles on Enterprise.⁠Source 20, Source 21, Source 22
  • You want published prices and vendor support: Cloud Starter is $20 a month billed annually; Enterprise, priced through sales, adds dedicated support.⁠Source 23

Questions buyers ask

Is n8n an alternative to building it yourself?

For building agents and workflows, it can be: n8n says you build them in n8n, visually or with code, on n8n Cloud or self-hosted.⁠Source 5, Source 6 Its Agents feature is in preview, and governing agents built outside n8n is not publicly documented.⁠Source 7

Does n8n support MCP and A2A?

MCP, yes: n8n can act as an MCP client, and as an MCP server through its MCP Server Trigger node, over SSE or streamable HTTP but not stdio.⁠Source 10, Source 11 First-party A2A support is not publicly documented.

Can n8n be self-hosted?

Yes. Self-hosted n8n can run on your own infrastructure, on-premises, or in a private cloud; without a license key, it runs as the free Community edition.⁠Source 24, Source 37 Air-gapped installs use a license certificate; otherwise Business and Enterprise license keys must ping n8n’s license server daily.⁠Source 23, Source 39

How do the costs compare?

n8n prices paid plans by monthly workflow executions; Cloud Starter is $20 a month billed annually, and one agent turn (preview) counts as one execution.⁠Source 7, Source 23 For DIY, the protocol specifications are openly licensed; build and running costs are not publicly documented.⁠Source 2, Source 3

How we compare

Read the full method

Every claim on this page links to a public source. Where none answers a question, the page says so.

We re-check every fact at least every 90 days. This page was last checked .

Something wrong or out of date? Tell us and we’ll correct it.

Sources

49 public sources, each with the date we checked it. Every one opens in a new tab.

  1. Source 1: google/adk-python README (GitHub) github.com · checked Back:abcdefg

  2. Source 2: Agent2Agent (A2A) Protocol Specification a2a-protocol.org · checked Back:abcdefghijklmnop

  3. Source 3: modelcontextprotocol/modelcontextprotocol LICENSE (GitHub) github.com · checked Back:abcd

  4. Source 4: Specification - Model Context Protocol 2026-07-28 modelcontextprotocol.io · checked Back:abcd

  5. Source 5: AI Workflow Automation Platform - n8n n8n · checked Back:abcdefghij

  6. Source 6: Choose how to use n8n n8n · checked Back:abcdefg

  7. Source 7: Build and manage agents n8n · checked Back:abcdefghijklmn

  8. Source 8: langchain-ai/langgraph README (GitHub) github.com · checked Back:abc

  9. Source 9: Designing MCP Gateway Uber's MCP Management Platform - Uber Blog uber.com · checked Back:abcdefghi

  10. Source 10: MCP Client Tool n8n · checked Back:abcdef

  11. Source 11: MCP Server Trigger n8n · checked Back:abcde

  12. Source 12: Integrating with Model Context Protocol (MCP) - Keycloak keycloak.org · checked Back to text

  13. Source 13: Agent Discovery - A2A Protocol a2a-protocol.org · checked Back:abcde

  14. Source 14: Enterprise Features - A2A Protocol a2a-protocol.org · checked Back:abcdefg

  15. Source 15: LangChain in n8n n8n · checked Back:abc

  16. Source 16: Building an MCP Ecosystem at Pinterest - Pinterest Engineering Blog medium.com · checked Back:abcdefgh

  17. Source 17: Security at n8n n8n · checked Back:abcd

  18. Source 18: The MCP Registry - Model Context Protocol modelcontextprotocol.io · checked Back:abcd

  19. Source 19: Authorization - Model Context Protocol specification 2026-07-28 modelcontextprotocol.io · checked Back:abcd

  20. Source 20: Set permissions and roles (RBAC) n8n · checked Back:abcde

  21. Source 21: Set up SAML n8n · checked Back:abcde

  22. Source 22: Set up OIDC n8n · checked Back:abcde

  23. Source 23: n8n Plans and Pricing n8n · checked Back:abcdefghijklmn

  24. Source 24: Compare editions n8n · checked Back:abcd

  25. Source 25: Introducing n8n Agents n8n · checked Back:ab

  26. Source 26: SDK Tiers - Model Context Protocol modelcontextprotocol.io · checked Back:ab

  27. Source 27: MCP Registry Aggregators - Model Context Protocol modelcontextprotocol.io · checked Back:ab

  28. Source 28: Streamable HTTP - Model Context Protocol specification 2026-07-28 modelcontextprotocol.io · checked Back:ab

  29. Source 29: Connect to n8n MCP server n8n · checked Back:abc

  30. Source 30: Stream logs to external systems n8n · checked Back:abc

  31. Source 31: n8n Trust Center n8n · checked Back:ab

  32. Source 32: Changelog n8n · checked Back to text

  33. Source 33: Release notes 1.x n8n · checked Back to text

  34. Source 34: Human-in-the-loop for tools n8n · checked Back to text

  35. Source 35: a2aproject/A2A README (GitHub) github.com · checked Back:ab

  36. Source 36: Transports - Model Context Protocol specification 2026-07-28 modelcontextprotocol.io · checked Back to text

  37. Source 37: Host n8n n8n · checked Back:abc

  38. Source 38: Try free then choose a plan n8n · checked Back to text

  39. Source 39: Manage your license n8n · checked Back:ab

  40. Source 40: A2A protocol roadmap a2a-protocol.org · checked Back to text

  41. Source 41: Integrate AI n8n · checked Back to text

  42. Source 42: n8n AI Terms n8n · checked Back to text

  43. Source 43: n8n AI agent builder n8n · checked Back to text

  44. Source 44: MCP servers n8n · checked Back to text

  45. Source 45: Use external secret stores n8n · checked Back to text

  46. Source 46: Your company's private network Blocks.ai · checked Back to text

  47. Source 47: Network requirements Blocks.ai · checked Back to text

  48. Source 48: Solutions: Agent sprawl Blocks.ai · checked Back to text

  49. Source 49: What is Blocks? Blocks.ai · checked Back to text