For the past few years, businesses have been obsessed with AI.
Almost every conversation about IT has focused on what AI can generate, integrate into, and how it works.
AI can do many things like draft emails, summarize meetings, create proposals, answer questions, produce images, and analyze data inside of spreadsheets.
Those capabilities are valuable, but they still require a person to initiate each task.
People are still the biggest resource in using AI. Humans provide the instructions, review responses, and decide what happens next.
And agentic AI changes that relationship.
Instead of simply responding to a prompt, an AI agent can be given a goal, determine the steps required, access approved tools and information, and complete work within defined boundaries – all within minutes.
This is where AI stops being something you occasionally consult and begins functioning as part of your business’s operation.
That shift is what led us to begin developing Navigate Compass, a service offering from our own effort to understand how AI would affect the managed IT industry, our clients, and the way we run Leverage IT Consulting.
We did not begin by asking, “What AI product should we sell?” Instead, we asked, “How is this technology going to change our business, and what do we need to do about it?”
An Idea Comes Back From Chicago
The story began when our founder and CEO, Eric Bayol, returned from a peer group meeting where the attendees walked through the evolution of artificial intelligence.
What they learned was how chat-based AI has evolved into something more capable like agentic AI.
These AI systems were not limited to answering questions or generating content rather that can reason through problems, use software tools, follow workflows, and take action. Large enterprises were already beginning to experiment with agentic AI and those capabilities were now starting to become accessible to small and mid-sized businesses, including managed service providers like us.
Our leadership team began wrestling with a question that many businesses are facing right now:
How will AI affect our industry, and how do we get ahead of it instead of getting run over by it?
We could already see that AI would change how IT services were delivered. It would influence how engineers solved problems, how information was documented, how customers received support, and how internal operations were managed.
The discussion became an EOS quarterly Rock, which is a defined, time-bound organizational priority with clear ownership.
Our Rock was straightforward to describe and much harder to execute: Find practical ways to use AI to improve the efficiency and quality of our service delivery. That assignment became my responsibility.
Testing the First Generation of AI Tools
I started by evaluating the AI products already available to managed IT providers.
There was no shortage of options with every software company adding an AI assistant, an automation feature, or an intelligence layer to its platform. I reviewed several solutions and narrowed the list down to the tools that appeared capable of creating real operational value.
All the technologies we reviewed delivered value, helping us automate tasks, improving documentation, and understanding where AI could support our team.
And they also revealed something larger: The more we used individual AI tools, the more obvious it became that the real opportunity was not simply adding an AI feature to each application. The bigger opportunity was creating an AI capability that could understand the business as a whole.
What if an AI system could know our processes, access our systems, retain organizational context, and complete work across multiple applications? That would be very different from buying another software license.
Discovering Autonomous Agents
Our next breakthrough did not come from a formal vendor demonstration, it came from our team experimenting with OpenClaw, an agentic tool designed to operate more like an autonomous assistant than a traditional chatbot.
We began exploring locally operated AI agents and quickly moved beyond basic testing, giving agent its own computer, email address, AI subscriptions, instructions, and defined workflows.
The agent was not simply sitting inside a chat window waiting for the next question; it could observe information, follow instructions, manage tasks, and complete work.
That distinction matters because a conventional AI tool might explain how to complete a task or produce a draft for you when an agent can be instructed to complete the process itself.
It gathers the required information, making decisions within established rules, interacting with systems, and returns the completed result for review. Our team saw this as a time multiplier and an expertise multiplier.
After experimenting and implementing it, our message was simple: Every business needs this.
From Personal Assistant to Business Engine
We learned that an agent could work as a personal assistant, handling real business operations when connecting to some of our systems.
We gave it access to approved documentation and taught it how Leverage IT operates, provides procedures, information, context, and examples of how our team works. Then our team assigned it a task that was real, tedious, and time-consuming: updating our internal procedures.
This was not a hypothetical demonstration designed to make technology look impressive; it was work that needed to be completed and that would normally require a substantial investment of time. Over a single weekend, the agent helped me refresh a body of procedures that would have taken me approximately two weeks to complete on my own.
That was the moment the business potential became undeniable – the agent was not producing generic text based on a vague request. It understood the systems we used, the processes we followed, and the operational context behind the procedures.
It was producing work at a level similar to someone actively working inside the business.
This was no longer about improving individual productivity by a few percentage points – it was fundamentally changing how work could be assigned and completed.
The Security Question
The more powerful the agent became, the more important the security questions became.
An AI system capable of accessing procedures, financial information, customer records, business applications, and internal conversations must be governed carefully. We began asking whether the technology could operate within a more controlled environment.
- Could we create an agent that ran locally or within the client’s own infrastructure?
- Could the organization maintain control over its information?
- Could we prevent sensitive company data from being transmitted to public AI services?
- Could we define what the agent was permitted to access and what actions it was authorized to take?
We began experimenting with local AI models and isolated environments. Even on limited hardware, the early results were strong enough to demonstrate that a secure and self-contained business deployment was not a fantasy. It was an engineering challenge that could be solved – that realization was critical.
Agentic AI would only become a legitimate business capability if it could be deployed with the same discipline organizations expect from their other technology systems. That means access controls, monitoring, logging, data governance, and clearly defined operational boundaries.
Power without governance creates risk. Power with governance creates leverage.
Two Perspectives Converge
Imagine giving every engineer access to the collective knowledge of the organization where a newer team member knew everything that a tenured veteran took years to learn. No more waiting for process documents and documentation – this “engineer” could raise the capability of the entire team.
As the team kept implementing and empowering the agentic AI tools, we kept returning to automation and business intelligence. You see, when an AI agent can securely connect to the systems a company uses to run its business, it gains a perspective that is difficult for any single employee to maintain.
It can examine revenue, expenses, sales pipeline activity, service performance, project status, and operational information. It identifies that revenue is trending below forecast while a major opportunity has stalled in the sales pipeline. It can recognize that project delays are beginning to affect costs and can surface patterns that might otherwise remain hidden until leadership reviews a monthly report.
That kind of cross-system awareness is difficult and time-consuming to assemble manually, and we learned that automation combined with deeper organizational intelligence was extremely valuable.
Agentic AI could become an operational capability, not simply another productivity application.
The Beginning of Navigate Compass: Our AI Service
Navigate Compass began taking shape through this internal experimentation, which was not created because we decided we needed another product to sell.
It emerged because we were testing technology against our own systems, workflows, security requirements, and business challenges. We discovered what worked, found where agents failed, and learned how much the quality of the underlying documentation matters.
We saw why access must be limited and governed, experienced the productivity gains firsthand, and began understand what it would take to make agentic AI useful, secure, and manageable for a small or mid-sized business.
Agentic AI is not a distant concept waiting to arrive at some point in the future – it’s already capable of supporting real operational work.
We’ve tested it thoroughly inside our own organization and found practical applications for procedure development, task automation, knowledge sharing, service delivery, and business intelligence.
The next question is the one that matters most to business leaders: What can agentic AI actually do inside your organization?
Leverage IT Consulting provides managed AI to businesses throughout Northern California and Northern
Click here to learn more about Navigate AI or contact us today.