Part 1: Tools: The Tech That’s Transforming SMB

By TechHouse Empower My Data (EMD) Team – Q3 2025

Why We Need to Pull Up and Check In on EMD for Our Organizations

It’s mid-2025, and Small and mid-sized businesses (SMBs) are getting access to tools that, until recently, only big enterprises could afford or manage. Microsoft has been at the forefront of innovations in AI (Artificial Intelligence) and low-code platforms. But with all the buzzwords – Copilot, GPT-5, Power Platform, Foundry, and various Data Platforms – it can be hard for a business leader to know what matters. What’s practical for a small company without any formal IT support? How about with a small IT department of 1 or 2 people? If your IT is outsourced, what should you be discussing with your provider?

Our Empower My Data team focuses on AI, Data, and App solutions. They wrote this article to share concrete trends and features from the last 3 months that can empower smaller organizations. We’ll explore:

  • Rapidly advancing technologies include Microsoft’s Power Platform (for building apps without coding), Microsoft 365 Copilot (an AI “assistant” across Office apps), and Azure AI Foundry (a marketplace of AI models like GPT-5). We’ll also explore how these affect our current understanding of data management and process workflows.
  • Realistic use cases and implementation paths: what you can do yourself vs. where you’d leverage an IT partner.
  • The human side: enabling your team, training, and adjusting processes to make these tools effective (and safe).
  • Actionable recommendations for business leaders to get started.

The Tools, Process, and People Framework

Tech is moving fast. Faster than ever. By far. It took 16 years for mobile phones to have 100 million users. It took only three months for ChatGPT to reach that worldwide. Small business leaders don’t need jargon when tech moves fast; they need clarity. We help our clients make sense of change by breaking it down into three simple parts: tools, processes, and people. This framework has been used for years across industries. It’s a practical way to understand how the moving levers in your business work and how to help it adapt. Tools are the apps, platforms, and devices we use daily, like email, accounting software, or AI-powered assistants. Process is how we get things done: the steps, systems, and routines that keep the business running. And people are the ones who make it all work: your team, your customers, your partners. When these three are in sync, businesses don’t just keep up, they get ahead.

Let’s dive into the top EMD trends under Tools, People, and Process.

I. Tools: The Tech That’s Transforming SMB

Microsoft’s summer 2025 releases have great potential for small businesses if used wisely. In this article, we will explore three of them: agents, Plan Designers, and Foundries.

1. Microsoft 365 Copilot’s New Agents (Researcher & Analyst)

Microsoft 365 Copilot has been around for several months, assisting with email drafting and document summarizing tasks. What’s new: in June-July 2025, Microsoft rolled out two advanced AI agents within Copilot: the Researcher agent and the Analyst agent. These aren’t separate products, but modes of Copilot that you can invoke for specialized help.

First, what is an agent? An AI Agent uses artificial intelligence to make decisions and perform tasks based on goals, context, and data.

It typically includes:

  • Perception: Understanding input like text, voice, data, etc.
  • Reasoning: Making decisions or drawing conclusions
  • Action: Executing tasks or generating responses
  • Learning: Improving over time through feedback or data

AI agents can be general purpose (like Copilot) or specialized (like the Researcher or Analyst agents).

  • The Researcher agent is like an on-demand business analyst or a researcher. It can comb through your organization’s data and pull relevant info from the web to answer a question or produce a briefing. For example, ask: “What are recent trends in online sales for businesses like ours, and how do our sales compare?” Researcher might scan your sales data (assuming it has access via, say, Excel spreadsheets in SharePoint Online) and also gather industry news or benchmarks from the web, then give you a summarized answer. This saves an employee hours of web searches and cross-referencing files. For a small business without a research department, that’s a big deal. We have often found market research to be one of the toughest capabilities to acquire, especially when developed in the context of our company’s strategic direction. It’s like suddenly having a librarian and analyst in one. Importantly, Copilot’s access to internal data is controlled by your permissions setup. It won’t show anything a user isn’t allowed to see (this is why proper setup is key, more on that later).
  • The Analyst agent is like having a strong data scientist on staff. It can interpret data, create calculations, and even write code (like Python or advanced Excel formulas) to analyze the problem it is solving. Let’s say you want to understand how your website traffic correlates to your overall business sales. You could provide a spreadsheet of website activity and your annual sales data export, and then ask, “Analyze our website traffic vs. sales over the past year and find any correlations”. The Analyst agent can run the numbers, generate charts, and explain findings in natural language. For SMBs, this bridges a talent gap: many can’t afford a full-time data analyst, but this tool can handle moderately complex analyses.

Why it matters: These Copilot agents democratize high-level knowledge work. Early adopters report massive time savings; what took days now takes hours or minutes. For example, a marketing manager could use Researcher to get a distilled report on competitors from internal memos and news articles combined. A finance lead could ask an analyst to identify expense anomalies across departments, which might usually require data export and pivot tables. The net effect is smarter decisions that are faster without hiring extra analysts. One user in a staffing firm context noted that M365 Copilot (with these agents) could summarize candidate resumes and client requirements side-by-side, speeding up placements.

Caveat: To get the most out of Copilot, it must be set up correctly. Permissions must be tuned so Copilot can see the relevant files (and not see what it shouldn’t). Microsoft provides a Copilot admin center for configuration. In a small biz with no IT admin, this is where an IT partner’s help is invaluable – we ensure, for instance, Copilot has access to SharePoint documents where policies are stored if you expect it to answer policy questions and prevent it from accessing a highly sensitive finance folder if that’s a concern. Additional security concerns are addressed below.

These agents are new and rolling out, so if you don’t see them yet, you might need to request access or updates.

2. Power Platform’s AI-Powered Plan Designer

The Power Platform (Power Apps, Power Automate, etc.) has long been Microsoft’s low-code toolkit, allowing creation of business apps and workflows with minimal coding. In July 2025, Microsoft introduced the Plan Designer in Power Apps, essentially “Copilot for solution building.” It lets you describe a business problem or process in plain language, and the AI will generate a proposed solution across the Power Platform.

For example, you could input: “We need an app for field technicians to log equipment inspections, a flow to notify the manager if any inspection is flagged as ‘failed’, and a dashboard to overview all upcoming inspections.” Plan Designer could create a prototype design that includes:

  • A series of Data Tables (maybe a list of equipment and inspection results).
  • A Power App form for technicians to submit inspections (with fields like equipment ID, status, and comments).
  • A Power Automate Flow that triggers on a “failed” inspection and sends an email to the manager.
  • A Power BI or Power Apps page showing scheduled vs completed inspections.

You or someone on your team could create a plan like this in minutes. It’s not magic. It uses templates and GPT-4/5 AI under the hood. To the business user, getting a jump-start like that feels magical. Microsoft’s documentation even suggests you attach existing assets (like an Excel document you use for inspections or a diagram of the process) to help the AI understand your needs.

Why it matters: Small businesses often have unique processes that off-the-shelf software doesn’t perfectly fit (or it’s too expensive). With Plan Designer, the barrier to attempting a custom solution is dramatically lower. You can brainstorm solutions on your own time, iterating many times as you explore your problem and how different solutions can provide different capabilities. If the problem calls for a simple solution, a tech-savvy employee could get a first draft app up and running in an afternoon. For example, you convert a simple store opening checklist into an app that can easily run on a phone. Plan Designer empowers the “citizen developer.”

However – and this is crucial – the output is a draft.  It is a first iteration that helps you explore the problem and possible solutions. If the problem is simple, you are confident you have gotten past the symptoms and can evaluate it for explicit and implicit errors, and it can run in isolation without interacting with your other software applications, then you may be able to move forward as is. We do encourage having an experienced professional take a look, especially if it were to malfunction which could cause harm to the organization.

If the solution is more complex or integrates with other applications of yours then engage an experienced guide to help you assess the design and ensure the application is helping and not harming. It is also a good idea to have an experienced objective person explore the problem with you to ensure you are looking at the root cause and not the symptom.

Addiontaly you will want an experienced eye on thing like security (did the app accidentally make data visible to all users when it should be restricted?) and efficiency (maybe the flow can be optimized). Are errors being captured, or are you receiving results that look accurate but are not? So, Plan Designer is best used as a collaboration tool between the business user and IT. You get to sculpt the clay; we help bake it in the oven, so to speak. In practice, a business leader might bring us what Plan Designer built and say, “Here’s what I got – can we polish this up and deploy it?” It’s a far easier starting point (and thus, lower cost) than a blank canvas.

3. Azure AI Foundry & GPT-5: Enterprise AI at SMB Fingertips

In August 2025, Microsoft’s Azure AI Foundry (launched earlier in preview) hit a milestone by adding the OpenAI GPT-5 model for general availability. Azure AI Foundry is a platform where you can find and use AI models for various tasks – imagine an app store for AI brains. It hosts both Microsoft’s own models and those from partners (including OpenAI).

GPT-5 is the successor to GPT-4 (the model behind much ChatGPT). It’s extremely powerful:

  • It can handle a 272,000 token context, meaning it can take in vast amounts of text in one go (roughly ~200 pages). For comparison, GPT-4 in ChatGPT handles about 8,000 tokens (6 to 7 pages) unless you have special versions. This means GPT-5 could, for instance, ingest your entire employee handbook and company policy documents at once and answer questions about them – no need to break it into chunks.
  • It has an “internal orchestrator” – essentially, it can break complex tasks into parts, use specialized sub-models to solve each, and then combine the answers. Think of it like an AI project manager delegating subtasks to expert mini-Ais. Some focus on math, some on language, etc. The result is better accuracy and fewer mistakes on complex queries.
  • It’s versatile: available in chat-oriented form, and faster but minor variants for quick tasks.

The key takeaway for small businesses is that the most advanced AI is no longer locked behind big-company labs; you can rent it. Azure AI Foundry lets you pay-per-use for GPT-5 and other models, with enterprise-level security. One immediate use case: if you often need to summarize or analyze very large documents  like contracts, reports, research papers relevant to your business, then GPT-5 can do it in one shot. Another is building a customer context chatbot that actually understands long and complex customer histories or product manuals.

As always AI generated content requires review so keep that in mind when creating automations that include AI generated content.

Now, using Foundry and GPT-5 requires far deeper technical skills than Copilot’s new agents or Plan Designer. It’s not an end-user tool; it’s more for your IT provider or a developer to integrate into an application or workflow. For example, we at TechHouse used Microsoft’s AI models to inspect and categorize hundreds of thousands of documents for one of our customers. Your team does not have to train the model. They can benefit from it.

Microsoft provides comparison tools in Foundry so we can test your specific task on multiple models. That helps us pick the right model. Maybe GPT-5 is overkill for some tasks where a smaller model is cheaper and just as good.

The bottom line: SMBs should be aware that if they have a problem that could be solved by AI (and many do, even if not obvious), the barrier to custom AI solutions is much lower now. You don’t need a data center or a huge budget. These models are cloud services. The main ingredient needed is expertise to implement them correctly for your use case. For most SMB business leaders, that means collaboration with an external partner like TechHouse. Your AI partner most likely brings both security and app development experience to the table. If your current IT provider is a more traditional MSP, then you want an AI partner like TechHouse with decades of experience partnering with other tech firms.

Fifteen years ago, small businesses couldn’t dream of custom AI; five years ago, maybe some dabbling with chatbots; now, in 2025, even a 50-person company can realistically have, say, an AI-driven analytics dashboard or an intelligent Q&A bot tuned to their data. We’ll discuss later how to approach such projects prudently.

Quick Tech Recap: In the tools section, we see a theme – AI is baked into everything Whether Copilot in Office, AI-assisting app building, or AI models for any task, it’s increasingly accessible. The message to a business leader: You don’t have to be a tech giant to leverage these innovations, but you do have to be strategic in choosing which to use and how.

In Parts 2 and 3 of this article series, we will explore how our approach to Processes and People will affect our strategy to respond to this rapid Tech Change.