We’re opening our Founding Design Partner Program to 10 engineering teams building AI agents, and I think the n8n community is one of the best places to find people pushing AI workflows into production.
As AI agents become capable of issuing refunds, updating CRMs, sending emails, accessing sensitive data, and executing workflows, we’ve found many teams eventually build their own approval workflows, authorization middleware, or policy checks to control what AI is allowed to do.
That’s why we built Globi Guard.
Globi Guard sits between AI agents and enterprise systems, intercepting every AI action before it executes. Teams define their own policies, and Globi Guard either allows the action, blocks it, or requires human approval.
We’re looking for 10 teams to work closely with us as Founding Design Partners.
You’ll receive:
• Free pilot access
• White glove onboarding
• Direct access to the founders
• Priority support
• A real opportunity to influence the roadmap
We’re intentionally keeping the group small so we can work closely with every team and build around real production feedback.
If you’re building AI workflows with n8n and think this could be valuable, we’d love to work with you.
Hi — this is closely aligned with our work on reliable n8n workflows, approval gates, idempotency, and audit evidence.
Our current evidence is self-directed sandbox/reference work rather than a live client production deployment, so we would not overstate that. We could contribute a bounded synthetic pilot around one workflow and one governed action, with deterministic allow/block/queue cases, retry checks, and a short findings report.
We would start without production data or credentials. Could you confirm whether this type of sandbox-stage team is eligible, the expected time commitment, and whether the pilot includes any case-study or publicity obligations?
I’m the lead architect at Terpek.ia. We are currently building and deploying autonomous WhatsApp AI agents for the healthcare/dental sector via n8n and OpenRouter.
The biggest bottleneck to scaling our deployments is exactly what you described: the liability of giving an LLM write-access to CRMs and booking calendars without a scalable authorization layer. Right now, we are hardcoding human-in-the-loop approvals and policy checks directly into n8n nodes, which is inefficient and hard to scale across multiple client instances.
Globi Guard fits perfectly into our architecture stack. We’d be highly interested in joining as a Founding Design Partner to integrate and stress-test your middleware in real-world, high-stakes environments (healthcare quoting and scheduling).
Hi, thank you for reaching out! We’d love to learn more about what you’re building. Could you please email me at jamilkweya@globiguard.com and cc my cofounder, gedeonmatabaro@globiguard.com? We’ll set up a call to learn more about your architecture, discuss your use case and questions, and explore how we can work together. Looking forward to connecting.
This is a really interesting idea. I think a lot of teams focus on building AI agents first and only realize later that they need proper approval and permission controls. Having a layer that can review actions before they happen makes a lot of sense, especially for production workflows. Wishing you the best with the design partner program!
I’m interested in joining the Globi Guard Founding Design Partner Program.
I’m an AI Automation Engineer working extensively with n8n, APIs, AI agents, RAG systems, CRMs, databases, and human-in-the-loop workflows. I’ve built production automations where AI agents generate and publish multilingual content, update WordPress and NocoDB records, process leads, access knowledge bases, and execute actions across external systems.
As these workflows become more autonomous, controlling which actions agents can perform, adding policy checks, and requiring human approval for sensitive actions are becoming increasingly important. Globi Guard sounds highly relevant to the systems I’m currently building.
I’d be happy to test it within real n8n workflows, provide detailed technical feedback, and work closely with your team during the pilot.
Hi, thank you for reaching out! We’d love to learn more about what you’re building. Could you please email me at jamilkweya@globiguard.com and cc my cofounder, gedeonmatabaro@globiguard.com? We’ll set up a call to learn more about your architecture, discuss your use case, and explore how we can work together.
Hi Beamlak, thank you for reaching out! Could you please email me at jamilkweya@globiguard.com and cc my cofounder, gedeonmatabaro@globiguard.com? We’d love to set up a call to learn more about what you are building and explore how we can work together. Looking forward to connecting! Thank you.
Most AI workflows eventually need a decision layer once they move beyond internal testing, especially when agents start interacting with CRMs, email platforms, or other production systems.
I’ve recently been building AI-powered n8n workflows and one challenge I keep thinking about is keeping business logic separate from authorization logic instead of embedding approval checks throughout every workflow.
A couple of questions:
• Does Globi Guard expose an n8n node or is it currently consumed through REST/webhooks?
• How much flexibility is there for custom approval policies across different business workflows?
I’d be interested in learning more about the Founding Design Partner program.
This is highly relevant to the type of AI automation systems we are building at Evozard Consulting Services, and we would be interested in joining the Founding Design Partner Program.
We work with n8n-based AI workflows that interact with CRMs, ERP systems, databases, email, documents, messaging platforms, external APIs, and approval-driven business processes. As these agents become more capable, controlling what they are allowed to execute becomes just as important as the intelligence behind them.
Globi Guard’s allow, block, or require-approval model aligns closely with our approach to production AI automation, particularly for:
CRM and ERP record creation or modification
Email and messaging actions
Financial or commercially sensitive workflows
Document generation and approval
Lead qualification and outreach
AI agents accessing operational data
Human-in-the-loop execution controls
Multi-tenant workflow environments
We are also developing a modular Sales Automation Platform with rule-based and AI-based lead evaluation, approval workflows, automation triggers, audit trails, usage controls, and isolated agent responsibilities. This could provide a practical environment to evaluate policy enforcement, approval routing, action-level authorization, and detailed execution logs.
We would be particularly interested in exploring:
How Globi Guard integrates with n8n workflows and custom API tools
Policy enforcement at workflow and action level
Multi-tenant policy isolation
Approval escalation and timeout handling
Audit logs and traceability
Protection against prompt injection or unauthorized tool usage
Rules based on user, workspace, action, data type, and risk level
Fail-safe behaviour when Globi Guard or an external system is unavailable
You can view our official n8n Creator profile and published workflows here:
We would be happy to share our architecture and real use cases, test the product within relevant workflows, and provide structured feedback from both technical and operational perspectives.
Please feel free to DM me with the next steps for the Founding Design Partner Program.
Regards,
Saumil Shah
Co-Founder
Evozard Consulting Services Private Limited
I can help you integrate Globi Guard into your AI workflows, ensuring every agent action is governed by secure approval policies while keeping automations scalable, compliant, and production-ready.
• Integrate with existing AI workflows
• Configure approval & policy rules
• Secure AI actions across enterprise systems
• Optimize governance with minimal overhead
• Support production deployment & scaling
Hi, I’m Haider, founder of DevTechSlopes and a full-stack/mobile engineer with 7+ years shipping production systems using React/Next.js, React Native, Node.js, .NET, and Azure. I’ve been building AI agent powered features recently, including Claude API/Anthropic SDK integrations, tool use and function calling, MCP, and RAG pipelines for client products in fintech, property management, and marketplace platforms. Given that our systems increasingly involve agents taking real actions like payments, data access, and workflow execution, a policy and approval layer like Globi Guard is exactly the kind of infrastructure we’d want to pilot. Happy to be considered as one of your Founding Design Partners. Let me know next steps.
I’d like to put my hand up for the design partner program, with one caveat stated upfront: I’m not an engineering team. I’m a solo automation developer building production workflows for small clients, mostly in n8n.
I mention it because the problem you’re describing is one I’ve already solved by hand, badly. In an invoice processing pipeline, anything the model couldn’t parse confidently went to a human review queue instead of being written to the ledger — a wrong figure entering accounting data silently is far more expensive than a slow process. In a multi-agent system I built with CrewAI, specialist crews sit behind human approval gates before anything executes. Both times I wrote the approval logic myself. Both times it was the least interesting part of the build, and the reason anyone was willing to trust the result.
So I’d be a small design partner, but a real one: production systems, actual clients, and a clear memory of what breaks when the authority layer is homemade.
If small-scale feedback is useful to you, I’d be glad to pilot it.
Yes, we have a native n8n node, so Globi Guard can be dropped directly into n8n workflows. We also expose APIs and webhooks, so teams can integrate however they prefer.
On approval policies, they’re fully configurable within the industries we currently support. Today that’s Finance, Accounting, Insurance, Healthcare, Legal, People Operations, Technology, and Customer Operations. Each domain has its own policy framework, approval logic, and governance rules.
We’re also working toward more generalized policy reasoning so organizations can define broader business instructions that aren’t tied to a specific industry while still keeping decisions predictable and auditable.
I’d love to learn more about the workflows you’re building in n8n as well. They sound very aligned with what we’re building.
Happy to set up a call and walk you through the Founding Design Partner Program if you’re interested.
Thanks Haider! We really appreciate your interest.
Your experience building production AI systems across fintech, property management and marketplaces sounds like a great fit for the Founding Design Partner Program.
We’d love to learn more about what you’re building and explore a pilot together. Please send me an email at jamilkweya@globiguard.com and cc gedeonmatabaro@globiguard.com so we can schedule a call and discuss the next steps.
Interesting idea. As more AI agents start handling real business tasks, having an approval layer before sensitive actions are executed makes a lot of sense. Looking forward to seeing how teams integrate this with existing n8n workflows. Good luck with the program!