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.

Alternatives to Build it yourself (DIY)

Build it yourself (DIY)

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

Summary

Build it yourself (DIY) is not a product: your own teams connect and govern agents with open protocols such as A2A and MCP, existing infrastructure, and open-source tools.⁠Source 1, Source 2, Source 3, Source 4 Uber and Pinterest each say they built their own MCP registry and access checks.⁠Source 5, Source 6

By Blocks.ai teamLast checked

Where DIY sits

Six layers of running AI agents at a company, and what DIY’s own public sources say it covers. Each alternative below gets the same six layers as a strip.

Public sources · checked 2 October 2026
  • You build it

DIY

  • Build agents: You build itOpen-source frameworks like ADK⁠Source 4
  • Host and run agents: You build itSelf-hosted, like LangGraph⁠Source 7
  • Identity and access: You build itYour own identity provider⁠Source 8
  • Registry and governance: You build itYour own agent registry⁠Source 9
  • Traffic between agents, tools, and models: You build itYour gateway and service mesh⁠Source 5
  • Agents across organizations: You build itA2A with API management⁠Source 10

How to read the strips

  1. 01Build agents
  2. 02Host and run agents
  3. 03Identity and access
  4. 04Registry and governance
  5. 05Traffic between agents, tools, and models
  6. 06Agents across organizations
  • Offered
  • Preview
  • Not included
  • Not publicly documented

Gateways, identity, and security

4 alternatives

Closest in kind to a DIY build, so listed first.

  • Cloudflare MCP server portals

    Cloudflare One feature that puts MCP servers behind one governed endpoint

    Where it sits⁠Source 11

    1. Build agents: Not publicly documented
    2. Host and run agents: Not publicly documented
    3. Identity and access: Offered, Identity provider or service token
    4. Registry and governance: Offered, Centrally managed MCP servers
    5. Traffic between agents, tools, and models: Offered, Proxy for MCP tool calls
    6. Agents across organizations: Not publicly documented

    A portal that puts multiple MCP servers behind one HTTP endpoint, with policies on who can connect and a log of tool requests.⁠Source 11 For portals, A2A support is not publicly documented.

    Cloudflare MCP server portals vs DIY

  • Kong AI Gateway

    Gateway that governs LLM, MCP, and agent-to-agent traffic

    Where it sits⁠Source 12, Source 13, Source 14

    1. Build agents: Not publicly documented
    2. Host and run agents: Not publicly documented
    3. Identity and access: Offered, Per-agent allow or deny lists
    4. Registry and governance: Preview, Konnect Catalog agents
    5. Traffic between agents, tools, and models: Offered, Gateway for LLM, MCP, A2A
    6. Agents across organizations: Not publicly documented

    A gateway for LLM, MCP, and A2A traffic; in 2.x, you run its data plane.⁠Source 15 Uber says it built its own MCP gateway, with a registry as the control plane.⁠Source 5

    DIY vs Kong AI Gateway

  • Okta for AI Agents

    Okta offering that gives AI agents a first-class identity so organizations can discover, onboard, protect, and govern them

    Where it sits⁠Source 16, Source 17, Source 18

    1. Build agents: Not publicly documented
    2. Host and run agents: Not publicly documented
    3. Identity and access: Offered, Agent identity and credentials
    4. Registry and governance: Offered, Universal Directory with human owners
    5. Traffic between agents, tools, and models: Preview, Agent Gateway for MCP tools
    6. Agents across organizations: Not publicly documented

    Okta says it gives AI agents a first-class identity.⁠Source 19 Admins register agents and set what each can access.⁠Source 20, Source 21 DIY builds can use an identity provider such as Keycloak as an MCP authorization server.⁠Source 8

    DIY vs Okta for AI Agents

  • Zenity AI Agent Security & Governance Platform

    Security and governance platform for AI agents, aimed at security teams

    Where it sits⁠Source 22, Source 23

    1. Build agents: Not publicly documented
    2. Host and run agents: Not publicly documented
    3. Identity and access: Not publicly documented
    4. Registry and governance: Offered, AI Observability agent inventory
    5. Traffic between agents, tools, and models: Offered, Tool-call blocking for coding agents
    6. Agents across organizations: Not publicly documented

    A SaaS security product.⁠Source 24 Zenity says it finds a company’s agents where they already run and can check their actions against the company’s rules.⁠Source 22, Source 25 Whether Zenity supports A2A isn’t publicly documented.

    DIY vs Zenity

Agent control planes

5 alternatives

  • Blocks.ai

    Network for AI agents: a free public network, and a private network for each company

    Where it sits⁠Source 26, Source 27, Source 28, Source 29, Source 30

    1. Build agents: Not included, Use LangChain or CrewAI
    2. Host and run agents: Not included, You run your agent
    3. Identity and access: Offered, Machine identity per agent
    4. Registry and governance: Offered, Private registry and Admin Console
    5. Traffic between agents, tools, and models: Offered, Private network for every agent
    6. Agents across organizations: Offered, Partners share only chosen agents

    Blocks.ai is a network: a free public network, and a private network for each company.⁠Source 29, Source 31 Blocks.ai doesn’t build or host agents.⁠Source 26, Source 27 Under DIY, frameworks such as LangGraph build and deploy them.⁠Source 7

    Blocks.ai vs DIY

  • Microsoft Agent 365

    Microsoft 365 control plane to discover, manage, govern, and secure AI agents

    Where it sits⁠Source 32, Source 33, Source 34, Source 35, Source 36

    1. Build agents: Not included, Use your chosen framework
    2. Host and run agents: Not included, You host your agent
    3. Identity and access: Offered, Entra Agent ID
    4. Registry and governance: Offered, Agent registry in admin center
    5. Traffic between agents, tools, and models: Preview, Tooling Gateway for MCP servers
    6. Agents across organizations: Not publicly documented

    A Microsoft 365 control plane for agents, with an agent registry and Entra agent identities.⁠Source 34, Source 35, Source 37 Its SDK doesn’t build or deploy agents; under DIY, frameworks such as Google’s Agent Development Kit do.⁠Source 4, Source 33

    DIY vs Microsoft Agent 365

  • MuleSoft Agent Fabric

    Control plane for agents, MCP servers, and APIs across platforms

    Where it sits⁠Source 38

    1. Build agents: Offered, Agent brokers in Agent Script
    2. Host and run agents: Offered, Agent brokers on CloudHub 2.0
    3. Identity and access: Offered, Omni Gateway authentication and authorization
    4. Registry and governance: Offered, Agent Registry in Anypoint Exchange
    5. Traffic between agents, tools, and models: Offered, Omni Gateway in request path
    6. Agents across organizations: Not publicly documented

    A control plane, hosted by MuleSoft, for agents, MCP servers, and APIs, with a registry scanners can fill and a gateway in the request path.⁠Source 38, Source 39, Source 40, Source 41 Uber says it built both for MCP servers.⁠Source 5

    DIY vs MuleSoft Agent Fabric

  • ServiceNow AI Control Tower

    What ServiceNow calls a central hub to discover, secure, govern, observe, and measure AI

    Where it sits⁠Source 42, Source 43, Source 44

    1. Build agents: Not publicly documented
    2. Host and run agents: Not publicly documented
    3. Identity and access: Offered, Scoped OAuth tokens (community docs)
    4. Registry and governance: Offered, AI inventory tied to CMDB
    5. Traffic between agents, tools, and models: Offered, AI Gateway (community docs)
    6. Agents across organizations: Not publicly documented

    ServiceNow describes it as a central hub to discover, secure, govern, observe, and measure AI.⁠Source 43 ServiceNow says it auto-discovers agents, models, and MCP servers into one inventory tied to the CMDB.⁠Source 43

    DIY vs ServiceNow AI Control Tower

  • Workday Agent System of Record

    Workday system of record to find, add, register, configure, monitor, and manage AI agents

    Where it sits⁠Source 45, Source 46, Source 47

    1. Build agents: Not publicly documented
    2. Host and run agents: Not publicly documented
    3. Identity and access: Offered, Agent System User per agent
    4. Registry and governance: Offered, Agent Registry in Management Hub
    5. Traffic between agents, tools, and models: Offered, Agent Gateway for Workday APIs
    6. Agents across organizations: Not publicly documented

    A functional area in each Workday tenant that registers agents built by Workday, partners, or you, each with a unique Workday identity.⁠Source 45, Source 46, Source 48 Agents built outside Workday currently register only through its API.⁠Source 49

    DIY vs Workday Agent System of Record

Platforms to build and run agents

6 alternatives

  • Amazon Bedrock AgentCore

    AWS platform for building, deploying, and operating AI agents

    Where it sits⁠Source 50, Source 51, Source 52, Source 53, Source 54

    1. Build agents: Offered, Harness managed agent loop
    2. Host and run agents: Offered, AgentCore Runtime
    3. Identity and access: Offered, AgentCore Identity workload identities
    4. Registry and governance: Offered, AWS Agent Registry with approvals
    5. Traffic between agents, tools, and models: Offered, AgentCore Gateway
    6. Agents across organizations: Offered, Registry shared through AWS RAM

    AWS’s managed platform to build, deploy, and operate agents, from modular services you can use together or independently.⁠Source 50 AgentCore runs in AWS Regions; DIY is assembled from parts such as Keycloak, Istio, and LangGraph.⁠Source 3, Source 7, Source 8, Source 55

    Amazon Bedrock AgentCore vs DIY

  • CrewAI AMP

    Platform for deploying, monitoring, and scaling CrewAI crews and agents

    Where it sits⁠Source 56, Source 57, Source 58, Source 59, Source 60, Source 61

    1. Build agents: Offered, Crew Studio visual editor
    2. Host and run agents: Offered, Managed infrastructure for crews
    3. Identity and access: Offered, Per-deployment allow lists
    4. Registry and governance: Offered, Agent Repositories
    5. Traffic between agents, tools, and models: Offered, Custom MCP server connections
    6. Agents across organizations: Preview, Public A2A agents

    CrewAI’s platform for building, deploying, and monitoring CrewAI crews and flows in production.⁠Source 57, Source 62 AMP’s docs don’t describe deploying other frameworks’ agents; with DIY, you pick your own frameworks, such as Google’s Agent Development Kit.⁠Source 4

    CrewAI AMP vs DIY

  • Gemini Enterprise Agent Platform

    Google Cloud platform to build, deploy, govern, and optimize AI agents

    Where it sits⁠Source 63, Source 64, Source 65, Source 66

    1. Build agents: Offered, Agent Studio low-code canvas
    2. Host and run agents: Offered, Agent Runtime
    3. Identity and access: Offered, SPIFFE-based Agent Identity
    4. Registry and governance: Offered, Agent Registry
    5. Traffic between agents, tools, and models: Offered, Agent Gateway
    6. Agents across organizations: Offered, Gateway calls to outside agents

    Google Cloud’s platform to build, deploy, govern, and optimize agents; its Agent Registry and Agent Gateway are Google Cloud services.⁠Source 63, Source 66, Source 67, Source 68 Pinterest says it optimized for MCP servers in its internal cloud.⁠Source 6

    DIY vs Gemini Enterprise Agent Platform

  • IBM watsonx Orchestrate

    Agent management platform to build, deploy, orchestrate, and govern AI agents

    Where it sits⁠Source 69, Source 70, Source 71, Source 72, Source 73

    1. Build agents: Offered, Visual builder and ADK
    2. Host and run agents: Offered, Agents run on watsonx Orchestrate
    3. Identity and access: Preview, Agent identity
    4. Registry and governance: Offered, Agentic Control Plane
    5. Traffic between agents, tools, and models: Offered, A2A calls to agent endpoints
    6. Agents across organizations: Offered, Partner A2A agents in catalog

    IBM calls it an agent management platform to build, deploy, orchestrate, and govern agents; those configured with its Agent Development Kit run on it.⁠Source 70, Source 74 IBM says its control plane also governs agents built elsewhere.⁠Source 72

    DIY vs IBM watsonx Orchestrate

  • LangSmith

    Platform for observing, evaluating, and deploying agents

    Where it sits⁠Source 75, Source 76, Source 77, Source 78, Source 79

    1. Build agents: Offered, Fleet agent builder
    2. Host and run agents: Offered, LangSmith Deployment runtime
    3. Identity and access: Offered, Fleet per-agent permissions
    4. Registry and governance: Offered, Centralized registry in Deployment
    5. Traffic between agents, tools, and models: Offered, Fleet forwards MCP tool calls
    6. Agents across organizations: Not publicly documented

    LangChain describes it as a platform for observing, evaluating, and deploying agents.⁠Source 80 Fleet creates agents without code.⁠Source 75 Agents can also run wherever you already run apps and send it traces.⁠Source 81

    DIY vs LangSmith

  • n8n

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

    Where it sits⁠Source 82, Source 83, Source 84, Source 85

    1. Build agents: Offered, LangChain-based AI Agent node
    2. Host and run agents: Offered, Your infrastructure or n8n’s
    3. Identity and access: Not publicly documented
    4. Registry and governance: Preview, Agents stored in projects
    5. Traffic between agents, tools, and models: Offered, MCP client for external tools
    6. Agents across organizations: Not publicly documented

    n8n says you build AI agents and workflows in it, visually or with code, on n8n Cloud or self-hosted.⁠Source 83, Source 86 Its Agents feature is in preview.⁠Source 84 Governing agents built outside n8n is not publicly documented.

    DIY vs n8n

Questions buyers ask

Does a DIY build keep a registry of every agent?

Only if you build or run one: A2A’s current specification prescribes no standard API for curated registries, and the official MCP Registry, in preview, does not support private servers.⁠Source 9, Source 87 Uber and Pinterest each say they built one for their MCP servers.⁠Source 5, Source 6

What does building it yourself cost?

A2A and MCP are openly licensed specifications.⁠Source 1, Source 2 The cost of building and operating your own registry, gateway, or control plane is not publicly documented. A2A’s docs say a curated registry means deploying and maintaining a registry service.⁠Source 9

Can a DIY build manage agents from other frameworks and vendors?

A2A says it gives agents built with different frameworks or languages, or by different vendors, a common language.⁠Source 1 Managing them stays with you: MCP says it can’t enforce its security principles at the protocol level, and A2A calls authorization logic implementation-specific.⁠Source 1, Source 88

When is building it yourself the right choice?

When you want no agent platform contract: A2A and MCP are openly licensed specifications you can build on.⁠Source 1, Source 2 When you want access checks from systems you already run, as Uber and Pinterest say they do with internal access control and mesh identities.⁠Source 5, Source 6

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

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

  1. Source 1: Agent2Agent (A2A) Protocol Specification a2a-protocol.org · checked Back:abcde

  2. Source 2: modelcontextprotocol/modelcontextprotocol LICENSE (GitHub) github.com · checked Back:abc

  3. Source 3: Security - Istio concepts istio.io · checked Back:ab

  4. Source 4: google/adk-python README (GitHub) github.com · checked Back:abcd

  5. Source 5: Designing MCP Gateway Uber's MCP Management Platform - Uber Blog uber.com · checked Back:abcdef

  6. Source 6: Building an MCP Ecosystem at Pinterest - Pinterest Engineering Blog medium.com · checked Back:abcd

  7. Source 7: langchain-ai/langgraph README (GitHub) github.com · checked Back:abc

  8. Source 8: Integrating with Model Context Protocol (MCP) - Keycloak keycloak.org · checked Back:abc

  9. Source 9: Agent Discovery - A2A Protocol a2a-protocol.org · checked Back:abc

  10. Source 10: Enterprise Features - A2A Protocol a2a-protocol.org · checked Back to text

  11. Source 11: MCP server portals · Cloudflare One docs Cloudflare · checked Back:ab

  12. Source 12: AI Agents - Kong AI Gateway docs Kong · checked Back to text

  13. Source 13: Agents in Catalog - Kong Docs Kong · checked Back to text

  14. Source 14: Kong AI Gateway product page Kong · checked Back to text

  15. Source 15: AI Gateway architecture - Kong Docs Kong · checked Back to text

  16. Source 16: Add AI agents manually (Okta Help Center) Okta · checked Back to text

  17. Source 17: Okta for AI Agents (product page) Okta · checked Back to text

  18. Source 18: Agent Gateway (Okta Help Center) Okta · checked Back to text

  19. Source 19: Okta brings first-class identity to AI agents with Agent SSO Okta · checked Back to text

  20. Source 20: Add and register AI agents (Okta Help Center) Okta · checked Back to text

  21. Source 21: AI agent resource connections (Okta Help Center) Okta · checked Back to text

  22. Source 22: AI Observability | See Every Agent, Know What it Touches | Zenity Zenity · checked Back:ab

  23. Source 23: Coding and Personal Agents | Zenity Zenity · checked Back to text

  24. Source 24: AWS Marketplace: Zenity Zenity · checked Back to text

  25. Source 25: Runtime Boundaries | The Runtime Boundary for Autonomous AI | Zenity Zenity · checked Back to text

  26. Source 26: Why Blocks? Blocks.ai · checked Back:ab

  27. Source 27: What is Blocks? Blocks.ai · checked Back:ab

  28. Source 28: Authentication reference Blocks.ai · checked Back to text

  29. Source 29: Your company's private network Blocks.ai · checked Back:abc

  30. Source 30: Joining a Blocks network Blocks.ai · checked Back to text

  31. Source 31: Pricing Blocks.ai · checked Back:ab

  32. Source 32: Agent 365 SDK frequently asked questions | Microsoft Learn Microsoft · checked Back to text

  33. Source 33: Microsoft Agent 365 SDK overview | Microsoft Learn Microsoft · checked Back:ab

  34. Source 34: Agent 365 identity | Microsoft Learn Microsoft · checked Back:ab

  35. Source 35: Agent Registry in Microsoft 365 admin center - Microsoft 365 admin | Microsoft Learn Microsoft · checked Back:ab

  36. Source 36: Bring your own (BYO) MCP server in Microsoft 365 admin center - Microsoft 365 admin | Microsoft Learn Microsoft · checked Back to text

  37. Source 37: Microsoft Agent 365 - Service Descriptions | Microsoft Learn Microsoft · checked Back to text

  38. Source 38: Agent Fabric Overview Salesforce · checked Back:ab

  39. Source 39: Get Started with Agent Fabric Salesforce · checked Back to text

  40. Source 40: Salesforce Hyperforce Overview Salesforce · checked Back to text

  41. Source 41: Agent Fabric Release Notes Salesforce · checked Back to text

  42. Source 42: What's new in AI Gateway v3.4 - September 2026 release (ServiceNow Community, AI Control Tower articles) ServiceNow · checked Back to text

  43. Source 43: AI Control Tower - ServiceNow (product page) ServiceNow · checked Back:abc

  44. Source 44: AI Gateway Implementation Guide (ServiceNow Community, AI Control Tower articles) ServiceNow · checked Back to text

  45. Source 45: Concept: Agent Security (Workday Administrator Guide) Workday · checked Back:ab

  46. Source 46: About Workday Agents (Workday Administrator Guide) Workday · checked Back:ab

  47. Source 47: Concept: Workday Agent Gateway (Workday Administrator Guide) Workday · checked Back to text

  48. Source 48: Set Up Agent System of Record (Workday Administrator Guide) Workday · checked Back to text

  49. Source 49: Register External Agents (Workday Administrator Guide) Workday · checked Back to text

  50. Source 50: Overview - Amazon Bedrock AgentCore (Developer Guide) AWS · checked Back:ab

  51. Source 51: Provide identity and credential management for agent applications with Amazon Bedrock AgentCore Identity AWS · checked Back to text

  52. Source 52: AWS Agent Registry: Discover and manage agents, tools, and resources AWS · checked Back to text

  53. Source 53: HTTP passthrough targets - AgentCore Gateway AWS · checked Back to text

  54. Source 54: Sharing a registry across accounts with AWS RAM AWS · checked Back to text

  55. Source 55: Amazon Bedrock AgentCore FAQs AWS · checked Back to text

  56. Source 56: Crew Studio CrewAI · checked Back to text

  57. Source 57: CrewAI AMP (platform docs introduction) CrewAI · checked Back:ab

  58. Source 58: Role-Based Access Control (RBAC) CrewAI · checked Back to text

  59. Source 59: Agent Repositories CrewAI · checked Back to text

  60. Source 60: Custom MCP Servers CrewAI · checked Back to text

  61. Source 61: A2A on AMP CrewAI · checked Back to text

  62. Source 62: Prepare for Deployment CrewAI · checked Back to text

  63. Source 63: Agent Platform overview Google · checked Back:ab

  64. Source 64: Agent Gateway overview Google · checked Back to text

  65. Source 65: Agent Identity overview Google · checked Back to text

  66. Source 66: Agent Registry overview Google · checked Back:ab

  67. Source 67: Set up Agent Registry Google · checked Back to text

  68. Source 68: Import A2A agents from Agent Registry Google · checked Back to text

  69. Source 69: AI Agent Builder | IBM watsonx Orchestrate IBM · checked Back to text

  70. Source 70: Welcome to IBM watsonx Orchestrate Agent Development Kit IBM · checked Back:ab

  71. Source 71: Agent identity overview IBM · checked Back to text

  72. Source 72: AI Agent Control Plane | IBM watsonx Orchestrate IBM · checked Back:ab

  73. Source 73: Partner A2A (Agent2Agent) agents IBM · checked Back to text

  74. Source 74: IBM watsonx Orchestrate IBM · checked Back to text

  75. Source 75: No-code agents with LangSmith Fleet LangChain · checked Back:ab

  76. Source 76: LangSmith Deployment LangChain · checked Back to text

  77. Source 77: Agent platform comparison LangChain · checked Back to text

  78. Source 78: LangSmith Deployment LangChain · checked Back to text

  79. Source 79: Remote MCP servers LangChain · checked Back to text

  80. Source 80: LangSmith: The Agent Engineering Platform LangChain · checked Back to text

  81. Source 81: Set up LangSmith LangChain · checked Back to text

  82. Source 82: LangChain in n8n n8n · checked Back to text

  83. Source 83: AI Workflow Automation Platform - n8n n8n · checked Back:ab

  84. Source 84: Build and manage agents n8n · checked Back:ab

  85. Source 85: MCP Client Tool n8n · checked Back to text

  86. Source 86: Choose how to use n8n n8n · checked Back to text

  87. Source 87: The MCP Registry - Model Context Protocol modelcontextprotocol.io · checked Back to text

  88. Source 88: Specification - Model Context Protocol 2026-07-28 modelcontextprotocol.io · checked Back to text

  89. Source 89: Network requirements Blocks.ai · checked Back to text

  90. Source 90: Solutions: Agent sprawl Blocks.ai · checked Back to text

  91. Source 91: Solutions: Partner networks Blocks.ai · checked Back to text