Where PlanVault™ fits

This page explains how PlanVault™ sits alongside common stack layers—without claiming to replace specific vendor products or open-source libraries. The focus is the governed API and event execution boundary.

Head-to-head

Where PlanVault™ sits next to your agent stack

PlanVault™ is the governance and execution layer that connects to your agent stack — existing LangChain, LangGraph, CrewAI, AutoGen, DSPy, OpenAI agent API, or MCP agents can run behind a PlanVault™ tool boundary, with tool calls, secrets, approvals, and recovery handled outside the model and every run recorded for audit and replay.

PlanVault™ keeps tool calls, secrets, approvals, and recovery outside the model.

Tool routing

PlanVault™Governed control layer

Centroid-based DB routing; shortlist built before the LLM

OpenAI agent APIs

Managed by the relevant OpenAI API surface

Agent frameworks

Often implemented in application logic or through LLM/graph routing

Enterprise governance tools

Often policy, catalog, or app-level routing

Self-hosted · VPC-ready

PlanVault™Governed control layer
Self-hosted in customer VPC; local/private models via LiteLLM; restricted-network deployments require enterprise validation Security details
OpenAI agent APIs

External API dependency

Agent frameworks

Self-managed infra, cloud LLM

Enterprise governance tools

Often SaaS-oriented; options vary by vendor

Catalog scalability

PlanVault™Governed control layer

Dynamic routing from catalogs of thousands of tools; adaptive shortlist per step

OpenAI agent APIs

128 tools (Assistants API); context-bounded (Responses API)

Agent frameworks

Depends on agent design and available context

Enterprise governance tools

Often manually or policy-managed catalog

Crash recovery

PlanVault™Governed control layer
Event-sourced FSM, auto-recovery, idempotency keys Security details
OpenAI agent APIs

State and recovery depend on the provider-managed API surface

Agent frameworks

Checkpointing is possible; recovery semantics often stay app-owned

Enterprise governance tools

Often tied to the vendor conversation or workflow model

Large responses

PlanVault™Governed control layer

Schema flattening, JSONPath extraction, stdlib tools, depth truncation

OpenAI agent APIs

Depends on token/context limits and API pattern

Agent frameworks

Usually handled at the application layer

Enterprise governance tools

Depends on product focus and integrations

Market context

Three different problems, often confused under one name

The AI agent governance market in 2026 has split into three fundamentally different categories that are not direct competitors to each other.

Governance Overlays

Added on top of existing frameworks (LangChain, CrewAI, OpenAI SDK). Intercept actions after the agent has already decided what to do. Examples: Microsoft AGT, Lakera, NeMo Guardrails, Galileo.

Cloud-native Managed Runtimes

Managed agent execution from a cloud provider. Governance via cloud IAM and provider-native audit. Deployment, residency, and contractual terms vary by provider and configuration. Examples: Google Vertex AI Agent Engine, Amazon Bedrock Agents, Azure AI Foundry.

Governed Execution Runtimes

The execution environment itself. Governance is built into the execution pipeline—not added on top. Includes planning, execution FSM, per-tenant encryption, and customer-controlled deployment. Examples: PlanVault (purpose-built). Adjacent tools with partial overlap: Temporal (durable execution, no AI governance), Orkes Conductor (workflow runtime, no per-tenant controls).

Most media attention in 2026 focuses on Governance Overlays and Cloud-native Runtimes. PlanVault is a Governed Execution Runtime—deployed in a customer-controlled environment with deployment posture defined by the customer.

vs. Cloud-native Managed Runtimes

Customer-controlled deployment vs. cloud-managed execution

Vertex AI Agent Engine, Bedrock Agents, and Azure AI Foundry are managed runtimes operated by the cloud provider. They support CMEK for encryption, but deployment posture, data residency, and governance configuration depend on provider-specific settings and contracts. PlanVault runs entirely within your own infrastructure.

Capability
PlanVault™
Vertex AI Agent Engine
Bedrock Agents
Azure AI Foundry

Customer-controlled deployment

✅ Full VPC / on-prem

❌ GCP-managed

❌ AWS-managed

❌ Azure-managed

Per-tenant BYOK (customer-operated KMS)

✅ AES-256-GCM + KMS

⚠️ CMEK via Cloud KMS (keys managed in provider control plane)

⚠️ CMEK via Cloud KMS (keys managed in provider control plane)

⚠️ CMEK via Cloud KMS (keys managed in provider control plane)

Compliance documentation model

Customer-operated documentation

Provider-specific documentation varies

Provider-specific documentation varies

Provider-specific documentation varies

LLM provider independence

✅ LiteLLM proxy

❌ Gemini-primary

❌ Bedrock models

❌ Azure OpenAI

Prompt-safe secrets

OpenAPI tool import

✅ Any spec

❌ Custom integrations

❌ Custom integrations

❌ Custom integrations

Self-hosted deployment

✅ Self-hosted / VPC

HITL native FSM gates

❌ Callbacks only

❌ Callbacks only

❌ Callbacks only

vs. Governance Overlays

Governed execution runtime vs. overlay tools

Governance Overlays solve a different problem: protecting existing agent frameworks. They are not direct alternatives to PlanVault, but are often evaluated side by side. The key distinction: an overlay intercepts actions; a runtime governs all execution from start to finish.

Capability
PlanVault™
Microsoft AGT
Galileo
Lakera Guard
NeMo Guardrails

Execution runtime

✅ FSM + Pekko

❌ Overlay only

❌ Overlay only

❌ Overlay only

❌ Overlay only

LLM-generated plans

Crash recovery

Policy enforcement

✅ FSM gates, hard-deny

✅ YAML / OPA / Cedar

✅ Colang

Per-tenant encryption

✅ AES-256-GCM + KMS

Prompt-safe secrets

Residency / deployment control

Self-hosted / customer VPC

Deployment-dependent

VPC / on-prem options

Deployment-dependent

Self-hosted option

Self-hosted

✅ Full stack

✅ Container

✅ SaaS-primary

Certifications (SOC 2 / ISO)

❌ None yet

SOC 2

SOC 2

OWASP Agentic Top 10

Mapping documented

Published mapping

Partial

Partial

Partial

Framework integrations

Examples (open-source)

✅ 13+ adapters

✅ API

✅ LangChain

Open-source

✅ MIT

Partial

vs. Governed Execution Runtimes

Purpose-built AI governance vs. general workflow runtimes

Temporal and Orkes Conductor are mature, battle-tested workflow runtimes. They provide excellent execution durability and can be extended to support AI agent use cases. PlanVault's advantage is not that it does things Temporal cannot — it is that it ships months of AI-specific infrastructure out of the box: OpenAPI/MCP tool import, LLM context management, per-org key isolation, streaming plan UI, and native HITL approval flows.

Capability
PlanVault™
Temporal
Orkes Conductor

Customer-controlled deployment

✅ Full VPC / on-prem

✅ Self-hosted or Temporal Cloud

✅ Self-hosted or Orkes Cloud

Per-tenant envelope encryption (isolated key per org)

✅ AES-256-GCM, per-org key isolation + KMS

✅ Payload Codecs (worker-side encryption)

✅ Payload Codecs (worker-side, via Temporal SDK)

Built-in HITL approval UI + FSM gate

✅ Native FSM gates + approval UI (no code)

⚠️ Native Signals (primitive exists, build UI yourself)

⚠️ Native Signals (primitive exists, build UI yourself)

OpenAPI / MCP tool import

✅ Any spec, zero code

❌ Custom activity code required

❌ Custom worker code required

LLM-agnostic execution

✅ LiteLLM proxy

⚠️ Bring your own LLM calls

⚠️ Bring your own LLM calls

Audit trail & replay

✅ Full execution replay

⚠️ Workflow history only

⚠️ Workflow history only

Cost control & budget gates

✅ Per-run budgets

❌ Not supported

❌ Not supported

Multi-tenant org isolation

✅ Native org_id isolation

❌ Namespace-level only

❌ Namespace-level only

OSS community & ecosystem

⚠️ Early-stage community

✅ CNCF project, large OSS community

✅ Commercial support + partner ecosystem

SOC 2 Type II certification

⏳ In progress

✅ Temporal Cloud certified

✅ Orkes Cloud certified

General-purpose workflow support

❌ AI agents only

✅ Any workflow type

✅ Any workflow type

Plain comparison

Categories next to PlanVault™

Workflow runtimes

Workflow engines excel at human-oriented processes and SLAs; PlanVault™ adds a layer for AI-initiated actions: bounded tool catalogs, execution policy, and a complete event trail.

Agent orchestration frameworks

Libraries and frameworks help describe agent graphs and logic; PlanVault™ focuses on what happens after planning: validation, FSM execution, journaled transitions, and side-effect policy.

Enterprise governance systems

Compliance and IAM platforms set access rules for people and services; PlanVault™ adds a boundary for AI runs: plan validation, tool execution, and reviewable evidence.

Tool routing and retrieval middleware

Retrieval and routing narrow context for the model; PlanVault™ holds the execution boundary: who may call which API, with which secrets, and with what audit trail after the answer.

Durable execution systems

Systems that guarantee long-running steps often do not define per-AI-run API call policy, human approvals, and audit evidence—that is PlanVault™’s boundary.

AI runtime safety controls

Moderation and guardrails reduce unsafe model outputs; PlanVault™ governs what happens in your systems: calls, approvals, idempotency, and recovery.

Contrast: planning versus governed execution

  • An LLM or agent produces or proposes a plan of actions.
  • PlanVault™ validates the plan and executes through runtime and FSM—rather than trusting unconstrained API calls.
  • State transitions are journaled; side effects are policy-gated.
  • Recovery, audit, replay, cost visibility, retention, and user-data operations are runtime-boundary concerns—not only model text.

Limitations and transparency

PlanVault™ is designed for customer-controlled deployment, network integration, automated SBOM publication, security review, and penetration-test readiness—see the security overview and technical security documentation. However, we do not yet hold or publish SOC 2 or ISO certifications.

Ecosystem and community are early-stage, but the platform already includes runnable examples, integration paths, and implementation docs—start with the documentation.

The buyer category for a governed AI execution layer is still forming. PlanVault™ is currently in closed customer testing, and qualified teams can request a guided evaluation through request access.

We do not publish unverifiable named customers or unsupported production-scale claims; data handling expectations are described in the Privacy Policy. The comparisons on this page are category-level, not a legal assessment of named vendor products; deployment models and contractual terms may vary.