Putting AI Large Language Models to Work: An SMB’s Guide

A Practical Framework for Moving Beyond the Hype

By Jon Hoehne, CMIT Solutions

If you’re a business leader, you’re likely familiar with tools like ChatGPT. The technology behind them, Large Language Models (LLMs), has captured the world’s attention. It’s a fascinating opportunity, but how do you translate that into real-world business value?

Think of LLMs not as a magical “AI brain,” but as a powerful new type of tool that understands, generates, and reasons with language. In many ways, we’ve been here before. When the first iPhone was released, it was a revolutionary platform, but its full impact on business wasn’t immediately clear to everyone. It was only when an entire ecosystem of apps and integrations emerged that it became the indispensable tool it is today.

LLMs are on the same trajectory. The key is to adopt this foundational platform methodically. Instead of trying to boil the ocean, we recommend a simple “Crawl, Walk, Run” approach to integrate LLMs into your operations.

Phase 1: Crawl

The “Crawl” stage is about empowering your team with low-risk uses of LLMs to boost individual productivity. The goal is to get everyone comfortable interacting with this technology on tasks they already perform.

• Mastering the Assist: Encourage your team to use LLMs as a creative and administrative partner for non-sensitive tasks. This includes drafting social media posts, brainstorming marketing slogans, and refining public-facing documents.
• Leveraging Built-In Tools: Many software suites, like Microsoft 365 with its Copilot features, now have LLMs built directly into the applications you use every day, often operating within your existing security framework.
• The Focus: In this stage, you’re not changing what you do, but how fast you do it. The aim is efficiency and familiarization, while establishing clear policies on not using proprietary data in public tools.

Phase 2: Walk

Once your team is comfortable, the “Walk” stage focuses on applying LLMs to solve specific business problems and automate text-based workflows in a secure manner.

• Solve a Pain Point: Identify a process bottleneck that involves language. Is your customer service team answering the same questions repeatedly? An LLM-powered chatbot trained on your public FAQ and product manuals can provide instant answers.
• Automate Workflows: Move beyond individual tasks to connected processes. Use an LLM to analyze and summarize anonymized customer feedback from surveys or to automate the creation of first-draft reports from non-sensitive data.
• The Focus: This stage is about process improvement. It’s where LLMs start to work within your systems, not just for individual employees.

Phase 3: Run

The “Run” stage is where you leverage LLMs for strategic advantages.

• Unlock Your Data Securely: More power is unleashed when an LLM is connected to your company’s own data (project history, customer records, etc.). which is done exclusively within a private and secure AI tenant.
• Create a Knowledge Engine: Imagine any employee being able to ask complex questions (“What was our solution for the client who had an issue with X three years ago?”) and getting an instant, context-rich answer based on your company’s collective knowledge, without that query ever leaving your secure environment.
• The Focus: In the “Run” stage, you’re building a defensible competitive asset. You are using LLMs to unlock the immense value hidden within your own business information, safely.

How Can Your MSP Partner Help?

Navigating this journey, especially the technical and security aspects, is where a Managed Service Provider (MSP) becomes a critical partner. A common concern is data security. You are right to be wary of feeding proprietary company information or sensitive customer data into a public LLM on the internet.

This is where the concept of a secure AI tenant becomes critical. Think of it like your private cloud storage account; it’s a walled-off, secure environment where your data can be used by an LLM without being exposed to the public or used to train a model for others.

An MSP is essential in architecting and managing this private space throughout your journey. They provide technical expertise to integrate third-party AI tools safely and can setup the architecture for your future secure AI tenant.

The LLM revolution is here, but its adoption doesn’t have to be overwhelming. By taking a measured, secure, and phased approach, any business can move from simple experiments to building a truly intelligent operation.