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.

Amazon Bedrock AgentCore vs LangSmith

Amazon Bedrock AgentCore

AWS platform for building, deploying, and operating AI agents

LangSmith

Platform for observing, evaluating, and deploying agents

Short answer

Amazon Bedrock AgentCore is AWS’s set of modular services for building, deploying, and operating agents; LangChain describes LangSmith as its platform for observing, evaluating, and deploying them.⁠Source 1, Source 2, Source 3, Source 4 AgentCore is a managed service in AWS Regions; LangSmith Cloud deploys agents on Plus or above, and Enterprise adds BYOC, self-hosted, or hybrid options.⁠Source 1, Source 2, Source 3, Source 5, Source 6, Source 7, Source 8

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.

Public sources · checked 2 October 2026
  • Offered
  • Not publicly documented

Amazon Bedrock AgentCore

  • Build agents: OfferedHarness managed agent loop⁠Source 2
  • Host and run agents: OfferedAgentCore Runtime⁠Source 2
  • Identity and access: OfferedAgentCore Identity workload identities⁠Source 9
  • Registry and governance: OfferedAWS Agent Registry with approvals⁠Source 10
  • Traffic between agents, tools, and models: OfferedAgentCore Gateway⁠Source 11
  • Agents across organizations: OfferedRegistry shared through AWS RAM⁠Source 12

LangSmith

  • Build agents: OfferedFleet agent builder⁠Source 13
  • Host and run agents: OfferedLangSmith Deployment runtime⁠Source 3
  • Identity and access: OfferedFleet per-agent permissions⁠Source 14
  • Registry and governance: OfferedCentralized registry in Deployment⁠Source 15
  • Traffic between agents, tools, and models: OfferedFleet forwards MCP tool calls⁠Source 16
  • Agents across organizations: Not publicly documented

At a glance

TopicAmazon Bedrock AgentCoreLangSmith
Hosting optionsA managed service in AWS Regions, including AWS GovCloud (US-West).⁠Source 1, Source 2, Source 17 A self-hosted edition is not publicly documented.LangChain’s cloud, on GCP or AWS.⁠Source 5 On Enterprise: BYOC on AWS, or self-hosted or hybrid.⁠Source 6, Source 7, Source 8
Agents hosted elsewhereAWS Agent Registry, AgentCore Identity, and Observability can cover agents outside AgentCore Runtime; Observability needs extra telemetry setup.⁠Source 1, Source 18, Source 19Agents can run wherever you already run apps and send traces to LangSmith.⁠Source 6 Listing them in a registry is not publicly documented.
Identity and accessWorkload identities in AgentCore Identity.⁠Source 9 Calls to Runtime agents use IAM SigV4 by default, or JWTs from any OAuth 2.0 identity provider.⁠Source 20, Source 21LangSmith API keys or your own auth handlers on deployed agents.⁠Source 22 Enterprise adds role-based access control, and SAML SSO and SCIM on Cloud.⁠Source 23, Source 24
Pricing modelBy use, billed per service, with no upfront commitment or minimum fee.⁠Source 25 Registry has a monthly free tier.⁠Source 25A Developer plan with one free seat, Plus at $39 per seat per month, and Enterprise through sales.⁠Source 8 Usage is metered in LSUs at $1 each.⁠Source 8
Generally availableSince October 2025; AWS Agent Registry since August 2026.⁠Source 17LangChain announced Deployment, as LangGraph Platform, as GA on 14 May 2025, and Fleet, as Agent Builder, on 13 January 2026.⁠Source 15, Source 26, Source 27, Source 28

What each one is

Amazon Bedrock AgentCore

Amazon Bedrock AgentCore is AWS’s platform for building, deploying, and operating agents with any framework and model.⁠Source 2 Its modular services, usable together or independently, include a serverless Runtime, Gateway with Policy, Identity, Observability, and AWS Agent Registry.⁠Source 2, Source 9, Source 10, Source 29, Source 30

LangSmith

LangChain describes LangSmith as a framework-agnostic platform for observing, evaluating, and deploying agents.⁠Source 4 LangSmith Deployment runs agents in production, and LangSmith Fleet creates and manages agents without code; LangChain pitches Fleet to non-technical teams.⁠Source 3, Source 4, Source 13 Agents hosted elsewhere can send it traces.⁠Source 6

The differences that matter

  1. Where the platform runs

    Amazon Bedrock AgentCore

    AgentCore is a managed service in AWS Regions, with no infrastructure for you to manage.⁠Source 1, Source 2 A self-hosted edition is not publicly documented.

    LangSmith

    LangSmith runs in LangChain’s cloud on GCP or AWS, or, on Enterprise, as BYOC on AWS or self-hosted or hybrid.⁠Source 5, Source 6, Source 7, Source 8

    AgentCore Memory, Policy, and Evaluations keep data in the primary Region; cross-region inference may move prompts and results out.⁠Source 31 LangSmith Cloud stores the services it lists in your sign-up region.⁠Source 5

  2. Controlling agents and access

    Amazon Bedrock AgentCore

    Policy in AgentCore can check each request through a Gateway against policies, written in natural language or Cedar, before allowing tool access.⁠Source 29

    LangSmith

    Fleet agents can be set to pause for approval before specific actions, and per-agent permissions set who can clone, run, or edit each one.⁠Source 14, Source 32

    Policy covers traffic through an AgentCore Gateway.⁠Source 29 On LangSmith Deployment, authorization is handled by auth handlers the developer writes.⁠Source 22

  3. Building agents

    Amazon Bedrock AgentCore

    The managed harness defines an agent from a model, system prompt, and tools; teams can also write the loop in Python, for example with LangGraph.⁠Source 2, Source 33

    LangSmith

    Fleet builds agents without code, from a description or template, which LangChain pitches to non-technical teams; code-first agents can be deployed with the LangGraph CLI.⁠Source 3, Source 4, Source 13

    The models behind Fleet’s three managed tiers may change over time.⁠Source 34 AgentCore’s harness works with Amazon Bedrock, OpenAI, Google Gemini, and OpenAI-compatible providers.⁠Source 2

For security teams

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

TopicAmazon Bedrock AgentCoreLangSmith
Network exposureCallers in a VPC can reach AgentCore over AWS PrivateLink; AgentCore can reach VPC resources without exposing them publicly.⁠Source 35, Source 36The Deployment data plane polls the control plane for deployment updates; the control plane never connects to it directly.⁠Source 37
IdentityAgent identities are workload identities; each request is verified independently; works with identity providers such as Entra ID or Okta.⁠Source 1, Source 9, Source 38Hosted Agent Servers default to API keys; self-hosted ones have no default authentication.⁠Source 22 Cloud Enterprise adds SAML SSO and SCIM.⁠Source 24
Access changes and revocationRemove an account from a registry share, or add an explicit deny on Runtime, Gateway, or Memory.⁠Source 12, Source 39Fleet sharing is revocable anytime; admins can deactivate any member’s personal access token, effective within a minute.⁠Source 32, Source 40
Audit trailCloudTrail can log Gateway calls (data events are off by default) and logs Registry control-plane calls; Policy logs its decisions.⁠Source 29, Source 41, Source 42, Source 43Enterprise audit logs of admin and configuration actions, kept up to 400 days, can be pulled hourly into a SIEM.⁠Source 44
ComplianceHIPAA eligible; AWS lists it as FedRAMP (Class C and Class D), SOC 2, ISO 27001:2022, and CSA STAR compliant.⁠Source 45, Source 46LangChain lists SOC 2 Type II, ISO 27001:2022, GDPR, and HIPAA for LangSmith.⁠Source 47, Source 48

Full comparison

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

Amazon Bedrock AgentCore and LangSmith compared on 18 criteria
Amazon Bedrock AgentCoreLangSmith
What it is
What it is and who it’s forAWS platform for building, deploying, and operating agents with any framework and model, for taking agents from proof of concept to production.⁠Source 1, Source 2LangChain calls it a framework-agnostic platform for observing, evaluating, and deploying agents.⁠Source 4 Deployment runs agents; Fleet is for no-code agents.⁠Source 3, Source 13
MaturityGenerally available since October 2025.⁠Source 17 AWS Agent Registry entered preview in April 2026 and became generally available in August 2026.⁠Source 17LangChain announced LangGraph Platform, now Deployment, as GA on 14 May 2025.⁠Source 15, Source 26 It announced Agent Builder, now Fleet, as GA on 13 January 2026.⁠Source 27, Source 28
Control
Agent registry and discoveryAWS Agent Registry catalogs agents, MCP servers, tools, and skills, on AWS or elsewhere, behind an approval workflow.⁠Source 2, Source 10 Consumers see only approved records.⁠Source 49LangChain says Deployment manages agents through a centralized registry with versioning.⁠Source 15, Source 50 Listing agents hosted elsewhere is not publicly documented.
Identity and access controlWorkload identities in AgentCore Identity.⁠Source 9 Runtime calls use IAM SigV4 by default or JWTs from any OAuth 2.0 identity provider.⁠Source 20, Source 21API keys or your own auth handlers on Agent Server.⁠Source 22 Fleet agents can use fixed or the user’s credentials.⁠Source 51 Enterprise adds RBAC, plus SAML SSO on Cloud.⁠Source 23, Source 24
Ownership, policy, and revocationGateway policies limit an agent’s tools.⁠Source 29 Approval workflows and curator deprecation control registry discovery.⁠Source 10, Source 43 An owner field is not publicly documented.Fleet shows each agent’s owner, has per-agent Clone, Run, and Edit permissions revocable at any time, and supports spend limits.⁠Source 14, Source 28, Source 32, Source 52
Audit log and observabilityCloudTrail can log Gateway calls (data events are off by default) and logs Registry control-plane calls.⁠Source 41, Source 42, Source 43 Policy logs decisions.⁠Source 29Enterprise audit logs, kept up to 400 days.⁠Source 44 Each deployment gets a tracing project; Fleet traces include tool calls and outputs.⁠Source 32, Source 37
Connection
How agents connectGateway can front agents and MCP servers hosted elsewhere at an endpoint URL.⁠Source 11 Private VPC connectivity and PrivateLink are supported.⁠Source 35, Source 36Agent Server offers A2A and MCP endpoints.⁠Source 53, Source 54 On self-hosted LangSmith with Deployment, agents run behind an ingress you configure.⁠Source 55 Enterprise can use PrivateLink.⁠Source 5
Agents across organizationsA registry can be shared with other AWS accounts through AWS RAM; Runtime agents can be opened to other accounts with resource-based policies.⁠Source 12, Source 39Fleet can share agents with users or your workspace; Agent Server agents can talk to A2A agents.⁠Source 32, Source 53 A cross-organization access model is not publicly documented.
Protocol supportRuntime agents can serve HTTP, MCP, A2A, or AG-UI.⁠Source 56 Gateway acts as one MCP server over its MCP targets.⁠Source 57 Registry validates MCP and A2A records.⁠Source 43Agent Server supports A2A, with documented gaps, over JSON-RPC.⁠Source 53 It can expose agents as MCP tools; Fleet uses remote MCP servers.⁠Source 16, Source 54
Frameworks, models, and clouds supportedRuntime works with LangGraph, CrewAI, LlamaIndex, Google ADK, OpenAI Agents SDK, Strands Agents, or custom frameworks, and any model.⁠Source 2Deployment runs Google ADK, Claude Agent SDK, Strands, CrewAI, and AutoGen agents via a wrapper package or the LangGraph Functional API.⁠Source 58
Operations
Deployment options and data residencyAWS says cross-region inference can move Memory, Policy, and Evaluations prompts out of the primary Region; AgentCore may store content to improve your service.⁠Source 2, Source 31Cloud on GCP (US, EU, APAC) or AWS (US).⁠Source 5 On Enterprise: BYOC on AWS, or self-hosted or hybrid.⁠Source 6, Source 7, Source 8
Compliance attestationsAWS lists AgentCore as FedRAMP (Class C and Class D) compliant.⁠Source 45, Source 46 It is HIPAA eligible, and SOC 2, ISO 27001:2022, and CSA STAR compliant.⁠Source 17, Source 45LangChain’s Trust Center lists SOC 2 Type II, ISO 27001:2022, GDPR, and HIPAA.⁠Source 47 LangChain says its SOC 2 Type II report spans the entire LangSmith offering.⁠Source 47
Support and SLAAWS says the Amazon Bedrock SLA applies to AgentCore.⁠Source 1 Basic Support is included for all AWS customers.⁠Source 59Base, Standard, and Premium support plans.⁠Source 60 Reasonable commercial efforts toward 99.5% API uptime for SaaS, measured quarterly.⁠Source 60
Time and effort to get runningAn AWS account with credentials and IAM permissions.⁠Source 33 AWS says its CLI quickstart reaches a running agent in a few minutes.⁠Source 33Cloud deployment needs Plus or above and an API key; LangChain says you can start in minutes.⁠Source 15, Source 61 Self-hosting the control plane needs Kubernetes and Enterprise.⁠Source 3
Pricing model and public pricesConsumption-based per service, no minimum fees.⁠Source 25 Registry has a monthly free tier; Runtime v1 CPU is $0.0895 per vCPU-hour.⁠Source 25Developer: one free seat; Plus: $39 per seat a month; Enterprise: sales.⁠Source 8 Usage is metered in LSUs, $1 each.⁠Source 8 Plus includes a free small serverless deployment.⁠Source 8
Building
Agent building toolsA managed harness takes a model, system prompt, and tools; or you write the loop in Python with a framework such as Strands, LangGraph, or Google ADK.⁠Source 2, Source 33Fleet builds no-code agents from a description or template.⁠Source 13 Code-first agents can be deployed with the LangGraph CLI and app templates.⁠Source 3
Model accessAWS calls AgentCore model-agnostic, naming OpenAI, Gemini, Claude, Nova, Llama, and Mistral models, in or outside Amazon Bedrock.⁠Source 1LangChain says LangSmith works with closed or open models.⁠Source 4 Fleet’s managed picker has three tiers (Fast, Pro, Max), not custom models.⁠Source 34
Integrations and ecosystemGateway has 1-click integrations such as Salesforce, Slack, Jira, Asana, and Zendesk, and can import AWS Partner tools bought in AWS Marketplace.⁠Source 1, Source 62Fleet has built-in tools like Gmail and GitHub.⁠Source 34 LangChain says Fleet agents work in Slack, Teams, and Gmail.⁠Source 63 Tracing works with OpenAI, Anthropic, and more.⁠Source 64

Which to choose

Choose Amazon Bedrock AgentCore if

  • You want agents in AWS’s serverless runtime, where calls use IAM by default, built with LangGraph, CrewAI, Strands Agents, or others.⁠Source 2, Source 20
  • You want each request through a gateway checked against policies, written in natural language or Cedar, before tool access.⁠Source 29
  • You want an approved catalog of agents and MCP servers, including ones on premises or in other clouds, shareable across AWS accounts.⁠Source 2, Source 10, Source 12
  • You need agents in AWS GovCloud (US-West), where AgentCore is generally available.⁠Source 17

Choose LangSmith if

  • You need LangSmith’s control plane and Agent Servers in your own infrastructure, for full data residency or air-gapped use, on Enterprise.⁠Source 3
  • Non-technical teams should build agents without code, from a description or template; LangChain says Fleet agents work in Slack, Teams, and Gmail.⁠Source 13, Source 63
  • You want to trace agents wherever they run, from any OpenTelemetry-compatible app or through integrations such as OpenAI, Anthropic, and CrewAI.⁠Source 6, Source 64, Source 65
  • You want Fleet’s per-agent Clone, Run, and Edit permissions, and human approval before specific agent actions.⁠Source 14, Source 32

Questions buyers ask

Can Amazon Bedrock AgentCore run LangGraph agents?

Yes. AWS lists LangGraph among the frameworks AgentCore Runtime works with, alongside CrewAI, LlamaIndex, Google ADK, OpenAI Agents SDK, and Strands Agents.⁠Source 2 LangSmith Deployment also runs agents built with other frameworks, such as Google ADK, CrewAI, and Strands.⁠Source 58

How is each one priced?

AgentCore is consumption-based, with each service billed for what you use and no minimum fee.⁠Source 25 LangSmith has a Developer plan with one free seat, Plus at $39 per seat per month, and Enterprise through sales, with usage metered in LSUs.⁠Source 8

Do they support MCP and A2A?

Agents on AgentCore Runtime can serve MCP or A2A, Gateway acts as an MCP server, and the registry validates MCP and A2A records.⁠Source 43, Source 56, Source 57 LangSmith’s Agent Server supports A2A, with documented gaps, and can expose agents as MCP tools; Fleet uses remote MCP servers.⁠Source 16, Source 53, Source 54

Can either one connect agents across organizations?

An AWS Agent Registry can be shared with other AWS accounts through AWS RAM; accounts outside the owner’s AWS Organization must accept an invitation.⁠Source 12 Agents on LangSmith’s Agent Server can talk to other A2A-compatible agents, but a cross-organization access model is not publicly documented.⁠Source 53

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

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

  1. Source 1: Amazon Bedrock AgentCore FAQs AWS · checked Back:abcdefghijk

  2. Source 2: Overview - Amazon Bedrock AgentCore (Developer Guide) AWS · checked Back:abcdefghijklmnopqr

  3. Source 3: LangSmith Deployment LangChain · checked Back:abcdefghi

  4. Source 4: LangSmith: The Agent Engineering Platform LangChain · checked Back:abcdef

  5. Source 5: Cloud (SaaS) LangChain · checked Back:abcdef

  6. Source 6: Set up LangSmith LangChain · checked Back:abcdefgh

  7. Source 7: Bring Your Own Cloud (BYOC) LangChain · checked Back:abcd

  8. Source 8: LangSmith Plans and Pricing LangChain · checked Back:abcdefghij

  9. Source 9: Provide identity and credential management for agent applications with Amazon Bedrock AgentCore Identity AWS · checked Back:abcde

  10. Source 10: AWS Agent Registry: Discover and manage agents, tools, and resources AWS · checked Back:abcde

  11. Source 11: HTTP passthrough targets - AgentCore Gateway AWS · checked Back:ab

  12. Source 12: Sharing a registry across accounts with AWS RAM AWS · checked Back:abcde

  13. Source 13: No-code agents with LangSmith Fleet LangChain · checked Back:abcdef

  14. Source 14: Agent platform comparison LangChain · checked Back:abcd

  15. Source 15: LangSmith Deployment LangChain · checked Back:abcde

  16. Source 16: Remote MCP servers LangChain · checked Back:abc

  17. Source 17: Release notes - Amazon Bedrock AgentCore AWS · checked Back:abcdef

  18. Source 18: Features of AgentCore Identity AWS · checked Back to text

  19. Source 19: Add observability to your Amazon Bedrock AgentCore resources AWS · checked Back to text

  20. Source 20: Authenticate and authorize with Inbound Auth and Outbound Auth AWS · checked Back:abc

  21. Source 21: Configure inbound JWT authorizer AWS · checked Back:ab

  22. Source 22: Authentication & access control LangChain · checked Back:abcd

  23. Source 23: Role-based access control LangChain · checked Back:ab

  24. Source 24: Authentication methods LangChain · checked Back:abc

  25. Source 25: Amazon Bedrock AgentCore Pricing AWS · checked Back:abcde

  26. Source 26: LangGraph Platform is now Generally Available: Deploy & manage long-running, stateful Agents LangChain · checked Back:ab

  27. Source 27: Now GA: LangSmith Agent Builder LangChain · checked Back:ab

  28. Source 28: LangSmith Fleet changelog LangChain · checked Back:abc

  29. Source 29: Policy in Amazon Bedrock AgentCore: Control Agent Interactions AWS · checked Back:abcdefg

  30. Source 30: Observe your agent applications on Amazon Bedrock AgentCore Observability AWS · checked Back to text

  31. Source 31: Cross-region inference in AgentCore Memory, Policy in AgentCore, and AgentCore Evaluations AWS · checked Back:ab

  32. Source 32: Access & oversight LangChain · checked Back:abcdef

  33. Source 33: Get started with Amazon Bedrock AgentCore AWS · checked Back:abcd

  34. Source 34: Essentials LangChain · checked Back:abc

  35. Source 35: Use interface VPC endpoints (AWS PrivateLink) with Amazon Bedrock AgentCore AWS · checked Back:ab

  36. Source 36: Connect to private resources in your VPC using VPC Lattice AWS · checked Back:ab

  37. Source 37: LangSmith control plane LangChain · checked Back:ab

  38. Source 38: Overview of Amazon Bedrock AgentCore Identity AWS · checked Back to text

  39. Source 39: Resource-based policies for Amazon Bedrock AgentCore AWS · checked Back:ab

  40. Source 40: Overview (administration) LangChain · checked Back to text

  41. Source 41: Log Amazon Bedrock AgentCore Gateway API calls with CloudTrail AWS · checked Back:ab

  42. Source 42: Enable CloudTrail data event logging for Amazon Bedrock AgentCore Gateway resources - Amazon Bedrock AgentCore AWS · checked Back:ab

  43. Source 43: Key capabilities - AWS Agent Registry AWS · checked Back:abcde

  44. Source 44: Audit logs LangChain · checked Back:ab

  45. Source 45: Compliance validation for Amazon Bedrock AgentCore AWS · checked Back:abc

  46. Source 46: Federal Risk and Authorization Management Program (FedRAMP) - Services in Scope - Amazon Web Services AWS · checked Back:ab

  47. Source 47: Trust Center - LangChain LangChain · checked Back:abc

  48. Source 48: LangSmith shared responsibility model LangChain · checked Back to text

  49. Source 49: Concepts and terminology - AWS Agent Registry AWS · checked Back to text

  50. Source 50: Assistants LangChain · checked Back to text

  51. Source 51: Agent identity LangChain · checked Back to text

  52. Source 52: Manage workspace administration LangChain · checked Back to text

  53. Source 53: A2A endpoint in Agent Server LangChain · checked Back:abcde

  54. Source 54: MCP endpoint in Agent Server LangChain · checked Back:abc

  55. Source 55: Enable additional LangSmith features LangChain · checked Back to text

  56. Source 56: Understand the AgentCore Runtime service contract AWS · checked Back:ab

  57. Source 57: Supported targets for Amazon Bedrock AgentCore gateways AWS · checked Back:ab

  58. Source 58: Deploy other frameworks LangChain · checked Back:ab

  59. Source 59: Compare AWS Support plans AWS · checked Back to text

  60. Source 60: Support Plans LangChain · checked Back:ab

  61. Source 61: Deploy your app to cloud LangChain · checked Back to text

  62. Source 62: Amazon Bedrock AgentCore Gateway: A secure AI gateway for agents, tools, and models AWS · checked Back to text

  63. Source 63: LangSmith Fleet LangChain · checked Back:ab

  64. Source 64: LangSmith Observability LangChain · checked Back:ab

  65. Source 65: Trace with OpenTelemetry LangChain · checked Back:ab

  66. Source 66: Why Blocks? Blocks.ai · checked Back to text

  67. Source 67: What is Blocks? Blocks.ai · checked Back to text

  68. Source 68: Your company's private network Blocks.ai · checked Back to text

  69. Source 69: Network requirements Blocks.ai · checked Back to text

  70. Source 70: Solutions: Agent sprawl Blocks.ai · checked Back to text

  71. Source 71: Solutions: Partner networks Blocks.ai · checked Back to text

  72. Source 72: Pricing Blocks.ai · checked Back to text