Policy on every agent call
Scope and policy are set on every agent and MCP call, then logged. Not a node somebody remembered to add.
mcp:crm.read allow
mcp:billing.write deny
agent:ticket.create allow
logged · scope set at platform
Moving from n8n / Governance and compliance
Governing AI takes a rock solid foundation.
Are you sure n8n is it?
Governing AI requires audit trails, versioning, compliance and trust. It is why n8n customers move to Tray.
Thanks — someone from our team will be in touch soon.
Govern integrations and AI confidently, with Tray
Who can change what
Access control and roles are operated by Tray, so who can build, edit and publish is a setting your team changes.
role buildeditpublish
admin
builder
viewer
a setting · not a rebuild
Every change is a version
Workflow versioning and history, so what changed, when and by whom is a matter of record rather than a reconstruction.
v12 r.patel 2h
v11 j.kim 1d
v10 r.patel 3d
reconstruction needed none
Secrets stay out of the build
Authentications are held and managed by the platform, not pasted into a workflow, a script or a prompt.
in the workflow auth.salesforce
held by the platform ••••••••
pasted into a prompt never
A record of every run
What ran, what it reached and what triggered it, kept as a platform record and available to stream into your SIEM.
ran enrich-accounts
reached salesforce
triggered by schedule
your SIEM
Evidence a reviewer accepts
SOC 1 and SOC 2 Type 2 with annual penetration testing, HIPAA, GDPR and CCPA, US, EU or APAC residency, signed DPAs and a published subprocessor list.
SOC 1 Type 2
SOC 2 Type 2
pen test
signed DPA
subprocessors
residency US EU APAC
reviewer pack ready
Why enterprises choose Tray
What a reviewer asks, and how Tray answers
| What a reviewer wants to know | How it works on Tray.ai |
|---|---|
| Where is policy applied? | On every agent and MCP call, by the platform, rather than in a step someone added to a workflow. |
| Who has to remember to turn it on? | Nobody. Coverage does not depend on whoever built each workflow. |
| What does the record show? | Every call: what ran, what it reached and what triggered it, kept as a platform record and available to stream into your SIEM. |
| Who keeps the controls working? | Tray. They are set up, kept configured and evidenced for you, not by your team. |
| How is access removed? | Access is granted through managed authentications and roles, so revoking one removes the access it granted rather than leaving it embedded in whatever used it. |
Industry recognized
3× Visionary Gartner Magic Quadrant for iPaaS, 2024, 2025 and 2026 
7× Leader Nucleus Research iPaaS Value Matrix, seven consecutive years Pioneer Gartner Emerging Market Quadrant for No-Code Agent Builders, 2026 14 Hype Cycles Gartner inclusions in 2026, including Agentic AI and Agentic Automation Controls you never configure
Tray.ai is the AI orchestration platform. We run the controls, you do not configure them.
- Access control you do not maintain, so limiting who can edit what is a setting rather than a purchase.
- Audit-log streaming, into the SIEM you already run, with retention following the policy you already have.
- Workflow versioning, so what changed, when, and by whom is a matter of record.
- Scope and policy on every agent and MCP call, evaluated by the platform and kept as a platform record.
Audited and certified
The evidence a reviewer accepts
SOC 1
audited annually
SOC 2 Type 2
with penetration testing
HIPAA
PHI handling
GDPR
EU data residency
CCPA
US data residency
700+
connectors managed
across integration, automation, and agents
1T+
processes run per year
on the platform
100%
workflow execution uptime
over the trailing 90 days
US, EU, and APAC data residency · Annual penetration testing, tested by independent assessors, with a bug bounty programme alongside. Signed DPAs and a published subprocessor list are available. Uptime measured on status.tray.ai and publicly checkable there. As of August 2026.
Visit the trust centerKeep reading
BlogRisk and patchingWhen the next n8n CVE drops, who patches it?BlogOperational overheadWhat it costs to run n8n yourselfBlogScale and reliabilityWhy n8n drops records without an error, and what you own at volumeBlogGovernance and complianceWho owns your SOC 2 report when you self-host n8n?BlogMigrationIntroducing the n8n to Tray migration kit
More on moving from n8n:Risk and patchingOperational overheadScale and reliabilityMigrationStart from the top
Contents
01 02
02 04
03 06
04 08
05 10
tray.ai · 12 pages
The guide · 12 pages
Five signs you've outgrown n8n
Five signs, what each one costs, and where your setup stands.
Get the guideWhy enterprises love Tray
“With Tray, we're now building agents that can troubleshoot access issues, automate provisioning, and reduce manual load on our IT teams. With a more agile and cost-effective integration strategy, Yext is now well-equipped to drive innovation, improve operations, and scale automation at an enterprise level.”
Frequently asked questions
Does n8n have governance features? +
Yes. There is a Guardrails node that enforces policy on text, and per-tool approval that the workflow builder configures. Those are real features and they work. The structural question is where the control lives: inside a workflow, added and configured by whoever built it, rather than as a platform property applied to every call whether anyone remembered or not.
How is policy enforced on Tray.ai? +
On the call itself. Scope and policy sit on every agent and MCP call, and every call is kept as a platform record. Nothing depends on a builder having added the right node to the right workflow, which means coverage is not a function of who built what and when.
Are RBAC, SSO, and audit logs standard or tier-gated? +
The controls exist on both platforms. What differs is who operates them. On a self-hosted install you configure them, keep them configured, and produce the evidence yourself. On Tray that work is the platform’s, which is what a reviewer is actually asking about when they ask who is responsible.
Is there a SOC 2 report for the platform running my automations? +
Yes. Tray.ai is SOC 1 Type 2 and SOC 2 Type 2 audited, with annual penetration testing. The reports cover the platform your workflows actually run on, which is the scope a reviewer cares about. Current documentation is on the trust center.
What about an MCP server with authentication left off? +
n8n’s MCP Server Trigger documents authentication as optional, with a random URL path as the default protection. That is a documented design choice, not a flaw, and it makes for a fast start. It also means the security posture of an MCP endpoint depends on what the person who created it selected, which is a different assurance model from a platform where the policy is not optional.
Still governing it workflow by workflow?
Let's discuss your governance gaps.
Thanks — someone from our team will be in touch soon.
GARTNER is a registered trademark and service mark, HYPE CYCLE and MAGIC QUADRANT are registered trademarks of Gartner, Inc. and/or its affiliates in the U.S. and internationally and are used herein with permission. All rights reserved. Gartner does not endorse any vendor, product or service depicted in its research publications and does not advise technology users to select only those vendors with the highest ratings or other designation. Gartner research publications consist of the opinions of Gartner’s Research & Advisory organization and should not be construed as statements of fact. Gartner disclaims all warranties, expressed or implied, with respect to this research, including any warranties of merchantability or fitness for a particular purpose.