AI at Gardiner + Company: How We’re Using It and How You Can Too – Matt Gardiner, CPA, CFE, Partner
Recently we rolled out Claude, an AI assistant from Anthropic, across our firm — including integrations directly inside Excel and Word, where a lot of our staff's day-to-day work happens. Our team uses it to clean up and reconcile data, build and improve spreadsheets, draft and refine written work, sift through large volumes of data quickly, and much more. Offloading that kind of repetitive work frees up our staff to spend more of their time on the things that actually move the needle for clients. Our use of this technology is walled-off completely: any information used doesn’t train the underlying model, and isn't shared publicly. Every deliverable gets reviewed and thoroughly vetted by the professionals you already work with. AI is a tool, not a replacement for judgment.
For most organizations, getting started is the hardest part. The technology moves fast, the options are overwhelming, and it's tempting to either dive in without a plan or put it off indefinitely. Neither works well. And in the meantime, staff often fill the gap on their own — pasting work into free consumer models that have no privacy protections, no oversight, and no record of what was shared. The good news is that a simple, structured approach gets you most of the way there.
Before walking through the steps, a few principles to keep in mind throughout. Start with low-stakes tasks like drafting emails, summarizing documents, or cleaning up meeting notes. Treat AI output as a first draft, making sure to always review it. Don't paste sensitive information (member data, payroll, confidential financials) into free public tools, which often use your inputs for training. And spend time training your people; the tool is only as good as the person using it.
Three steps to get started:
- Put an AI Acceptable Use Policy in place. Before anyone touches a tool, write down the rules: which platforms are approved, what data can and can't go in, who reviews AI-assisted work, and what happens if someone misuses it. This is the foundation — don't skip it.
- Train your team and let them find the use cases. Do hands-on training, share examples of good prompts, and pick someone internally to be the go-to person. Then give your team room to experiment within the guardrails — the best applications almost never come from the top down. Management can't see every workflow, every recurring annoyance, every place a few minutes get lost each day; the people doing the work can. Run it for 60–90 days and have them flag what's working and what isn't.
- Review and expand. After the pilot, look at what you saved in time, what the output quality looked like, and where the policy needs updating. Then expand to more teams or use cases. This is an ongoing process, not a one-and-done.
If you want to talk through how AI might fit into your organization, or if you’re curious how we might be leveraging it in ours, feel free to reach out!