AI Technical Operations is Here

Introduction: Pioneering the Future of Day 2 Operations

Born out of a need to find a transformative way to automate infrastructure, at System Initiative, we've pioneered AI Technical Operations—a fundamentally new approach for running Day 2 operations. It's designed so that compliance is a native component of our day-to-day flow, rather than something that gets tacked on after the fact. Our centralized methodology, using System Initiative, manages all our infrastructure and our user cradle-to-grave Role-Based Access Control (RBAC) lifecycle operations, delivering operational efficiency that traditional approaches simply can't match.

It sounds biased, but using System Initiative for this has been an absolute breeze—it has transformed what was previously frustrating toil and operational overhead into a true competitive advantage.

User Operations - The Background

One of the many things we use System Initiative for is to help us manage our user accounts and role assignments across all the services we use and delegate access to. Here are some examples:

• AWS - Managing access to our AWS resources, accounts + group delegations
• Tailscale User accounts - VPN access, ACLs, and network permissions
• PostHog User accounts - Product analytics and user tracking
• Hubspot User allocation - Customer Relationship Management
• JSR.io - Package registry for JavaScript
• Ashby User accounts - Recruiting platform

.. and many others spanning other parts of our organisation.

Having a single unilaterally extensible auditable place for all data and operations related to user assignments makes it trivial to complete any compliance information request in this domain, including those that our lovely auditor hasn't requested yet.

As a simple demonstration of the data model, watch how quickly I can search and group through the model to find fully qualified component references for different service names:

The golden rule for us is - if the service has an automation-compatible interface, we include it within the scope of our Technical Operations Workspace to further hydrate the capabilities of this process for us.

Oh, and before you ask, yes - System Initiative supports integrating with anything that has an API or CLI interface. Because System Initiative welcomes user-authored integrations and custom code, we aren't forced to choose vendors that claim to cover every possible current and future need. When we need to move fast—like redeploying a Docker image to an unexpected new provider—we can, knowing our platform is there for us.

AI-Powered Operations with MCP

Now here's where things get really wild—our Model Context Protocol (MCP You can find the complete source code for our AI agent MCP server on GitHub.) server interfaces AI queries directly with the graph data model for our user operations. In the scope of Technical Operations, it allows us to do the following, all with natural language:

• Manage user permissions and access controls in the cradle-to-grave lifecycle (Think user onboarding/role changes/offboarding and the associated permission changes)
• Query when a user received a particular permission allocation and who granted it (Think off-hand auditor request)
• Cleanly roll out the new service permissions into our existing organisation (Think post proof of concept vendor rollout)

Think about the last point: how often does a company roll out a new internal service, give early access to just the launch team, and then—months later—you're still filing torturous tickets because access was never rolled out properly?

Demo Time

I've talked enough - Let's walk through a real-world example of how we handle user offboarding. When someone leaves our organization, we need to remove their access from a plethora of services quickly and safely.

The Traditional Approach vs. Our Approach

In traditional operations, removing a user would involve:

1. HR / People Ops issues a ticket to IT / DevOps Team / Platform Engineering, informing them to deactivate the user
2. A team member ack'ing and offboarding the user from each service individually by:
   • Finding the user account
   • Removing permissions manually
   • Hoping you didn't miss any services
3. Audit Trail? The poorly written IT ticket and an indecipherable comment trail

With System Initiative and our MCP integration, anyone in the organization can propose a change using natural language.

As an example, here's how we'll remove me from the System Initiative Organization (I'm a bit too big for my boots anyway):

1. AI Query: "Use the deletion Change Set to remove john@systeminit.com from all our services"
2. Change Set Creation: The MCP server uses a simulation/Change Set
3. Propose Changes: The MCP server systematically reads the unified data model and finds references connected to John, proposing them for deletion.
4. Human Review: The proposed changes are reviewed by an Administrator to ensure the intention is correct
5. Safe Execution: Once approved, all user assignment modifications are audited and executed atomically

This is what it looks like:

Video Demonstration

In this video demonstration, we show the complete process of using our MCP server to safely remove a user from all our external services, including the creation of a change set, the review process, and final execution.

Worthy Callouts:

1. Common language prompt: No specialist knowledge needed to propose the change - anyone in our organization can do it.
2. Deep data model interrogation: No need to remember where they might be referenced in a system I don't have visibility of, it grabs them all.
3. Human review and approval: A workspace approver (in our case, our Technical Operations team) is prompted in Slack to review and accept the change
4. A deep archival audit trail is maintained: We don't have to worry about needing to "create a ticket to maintain compliance", it's all done automatically in the System Initiative platform.

The Benefits We've Experienced

Using System Initiative for our own operations has given us:

• Operational efficiency: Complex multi-service operations become simple AI queries
• Safety: All changes go through safe simulations and our approval workflow with full visibility
• Audit Trail: Complete history of who changed what and when
• Product feedback: Real-world usage helps us improve the platform
• Team confidence: Everyone can safely manage operations, not just domain specialists

I truly believe this approach to operations management represents the future of infrastructure automation. By combining a deep data architecture model with AI-powered automation and human oversight, we've created a system that scales with our team while maintaining safety and compliance.

If you're interested in learning more about how System Initiative can transform your operations, join us on Discord or email me at john@systeminit.com, I'd love to talk to you about it.

Don't tell anyone though — my life has never been so easy.