Forbes

Should You Connect AI To Your Financial Data? It Depends. Here’s Why.

By October 15, 2026No Comments

(This column originally appeared in Forbes)

A common question I get asked is about AI security. Would you connect your company’s financial data to an AI chatbot like ChatGPT and Claude? Are you worried about the privacy and security of your data?

The answer: It depends on your appetite for risk.

AI Connectors For Your Data

Most of the well-known chatbots now have, out-of-the-box, plug-ins where you can connect data from QuickBooks, PayPal, Outlook, Dropbox, Gmail and countless other common business applications. Both OpenAI and Anthropic have released small business versions of their platforms — and training — that can not only access this information easily but also offer skills and workflows for small businesses to create agents around that data. If your business is using an accounting, CRM, HR or other system that isn’t on their list, don’t fear. It’s very possible that you can hire a developer to use an MCP (Model Context Protocol) connector to these systems.

Should you do this? Many of my clients — small business owners — are. In 2025 they were just using AI chatbots to get answers from the Internet. But this year they’re starting to get answers from their actual data. They’re writing prompts like:

  • Analyze our profitability by customer, product line, manufacturer, salesperson and branch over the past 24 months. Identify our 10 most and least profitable areas, explain what is driving the differences, and recommend five actions to improve gross margin.
  • Using our CRM pipeline, historical sales, inventory, purchasing and seasonality data, forecast demand for the next six months by major product category. Identify potential stockouts, excess inventory and purchasing changes we should make now.
  • Review our accounting, CRM and HR data and identify the 10 biggest opportunities to increase EBITDA over the next 12 months. Quantify the potential financial impact of each recommendation and rank them by impact, difficulty and time required to implement.

Of course, this is just scratching the surface. These small business owners are creating prompts tailored to their business and industry. You can too. You can make these platforms available for all of your employees which then sets them up for three big benefits.

The Rewards Of AI Chatbots

For starters, they can use AI to prompt on company data that will help them do their job better. Also, by using AI more as part of their jobs they’ll be less fearful that this technology will replace them. That’s because the great majority of my clients are not only not planning on replacing their workers with AI but are still desperately in need of talent. To them, AI can help their people be more productive, and this is a big first step in that direction.

Finally, getting your people to use an AI platform on your company’s data for prompting will set them up for using agents in 2027 or 2028. This year, agents have been introduced by software vendors and developers and most of my clients have been underwhelmed. Agents in 2026 are still in their early stages, untrained and (in my opinion) mostly unreliable to set upon the mundane tasks they’re designed to accomplish. But they’ll get better in the coming year or two and not only will vendors be pushing their use, a good number of employees who are now over the hump of their AI fears will not only be more open to this kind of automation but may come up with their own ways to use agentic AI.

The Risks Of AI Chatbots

It’s exciting. But is there a risk to connecting your data to these AI platforms? Oh yes.

Despite the promises, assurances and pledges made by the big tech companies please know that your data is neither 100% private or secure. You probably already know this. You frequently read of breaches occurring at companies and even the government, places you would think are locked down but are never completely. You also read the news reports of nefarious big tech using private data for nefarious purposes, whether by mistake or through greed. That’s never going to end.

In the end it’s about the risk you want to assume and the rewards you want to reap.

For example, OpenAI recently released a connector that can access my personal banking and investment information in order to provide me with financial advice. I’m not comfortable with the potential risk of this connector. I’m promised up, down and sideways that it’s “perfectly safe” but this is my money we’re talking about. And besides, I’m a Certified Public Accountant, so I’m pretty good with my personal finances. In other words, the rewards for doing this are not enough to justify the risks.

But it’s different for other parts of my life. In my business, I’ve connected ChatGPT to my accounting, CRM and other data because it provides a significant amount of productivity savings and insights for both myself and my team. I use ChatGPT to help me research and for editing. Personally, I share a lot of my eating, exercise and travel activities to get its help. Am I risking that this information could be compromised? Yes. But to me the rewards are greater than the risk I’m assuming. So I do it.

When we use AI our data is going to the cloud. It’s encrypted. It’s stored in data centers on Microsoft Azure, Google Cloud, AWS or a number of other cloud-based services. These are the same cloud-based services that we’ve been using over the past decade in both our professional and personal lives. Nothing’s really changed. Your competitors haven’t figure out how to access your pricing information or customer lists and I’m pretty sure that 1) they’re in no better position to do this just because you’re using an AI interface and 2) they probably don’t care as much about your little business as you think they do.

Your Choice: Risk Vs. Reward

It’s all risk and reward. You take the risk of purchasing a piece of equipment, investing in a property, hiring that great new salesperson or buying a competitor’s business because you’re hoping that the rewards will be substantial enough to justify. It’s the same with AI.

The AI “safety” and “security” question is like, all other questions, unresolvable. Nothing is 100% safe or secure. You take risks in everything you do in your life, from getting on an airplane to eating a Big Mac. But you gauge the reward from doing so and make your decisions based on that. It’s no different for any business decision.