A couple of colleagues of mine who own MSPs around the United States had a substantial lead getting AI to show business value. They dove in early and were extolling its virtues, how it was helping them analyze data in their businesses. I reluctantly dove into myself, mostly for basic stuff at first, editing photos to embarrass family members and play jokes on people. But eventually that morphed into asking AI questions about business, about analytics, about marketing. I dove in heavily on the marketing side once I saw how valuable AI was at creating marketing content, analyzing our website’s value proposition, and updating pages and layout, all without paying a web developer or a marketer. That was a huge win for the business.
The next piece was data. My process for pulling data used to take about ten to twenty hours a month, going through our professional services automation system, our accounting system, our satisfaction survey system, and every other system we work in. We’re an E.O.S. shop, that’s the Traction methodology by Gene Wickman, and it’s brilliant for small and medium-sized businesses building a cadence to operate and grow. Every week in our Level Ten meetings, we’d review scorecards, ticket volume, resolution time, all the numbers tied to customer satisfaction and business health. I had a spreadsheet where I calculated all of this before feeding it into our E.O.S. software, Strety.
One day I just uploaded that spreadsheet to Claude and asked if it could pull this data directly. Claude said yes, and walked me through creating a read-only API account with ConnectWise, our P.S.A. system, so it could only read data, never modify anything. About ten minutes and five short questions later, I had a fully functional, real-time dashboard. It even caught and corrected some numbers I’d miscalculated.
From there I kept refining it: removing unnecessary metrics, adding valuable new ones, and building filters for daily, weekly, monthly, quarterly, biannual, and annual views, all without writing a single line of code. I asked for tickets and projects to hyperlink back to ConnectWise, then realized it’d be even better as a pop-up showing all the ticket details right there.
That became the pattern: keep asking, what do I want to know about my business that I can’t easily get by digging through ConnectWise myself? One dashboard I built tracks every project and flags any that got pushed back, showing who pushed it back, their justification, and how many days it slipped from the original date. We aim to complete most projects within thirty days. That visualization revealed a real process gap: we were scheduling tickets with customers but not entire projects, phase by phase. We’ve since changed how we schedule and coordinate project work because of it.
I’ve now built more than a dozen dashboards. At first they ran just for me, on an MCP server on my MacBook Pro. Then I wanted my team to have access too, so I built an emailing function, scheduled PDF and dynamic html reports going out daily and weekly, integrated directly with Microsoft three sixty-five. Claude built all of it. I didn’t write a line of code.
Next, I moved everything off my MacBook onto a dedicated Windows eleven machine in the office, an old computer that had just been sitting around. I asked Claude to package everything up for a Windows installation, and it did, though it took about an hour of back and forth catching path errors between Mac and Windows before everything ran cleanly.
Some dashboards, like the ones pulling QuickBooks Online financials or quoting data, aren’t meant for my technicians to see. So I had Claude build an authentication system using Microsoft Office, with granular permissions per dashboard, all through a central landing page which Claude also built. Now I can send invites to give technicians access to specific dashboards based on need.
The result has been huge. We’ve eliminated most of our daily stand-up meetings because everything that needs attention is now visible in the dashboards in real time. That’s freed up at least ten hours a week across the team. We’re now seriously considering whether we even still need Strety, since our custom dashboard may soon do everything it does, and more, fully tailored to us. I expect within a week or two, we’ll have replaced it entirely. To be clear, Strety is a great product, and if you’re not building your own system, you should absolutely use it. It just happens that in our case, we’re on track to replace it.
One important note: once these dashboards are built, they don’t consume Claude tokens to run. They pull and manipulate data directly through code Claude wrote. I only use tokens when I’m asking Claude to build or modify something, not every time a dashboard refreshes.
Once you’ve got that foundation, scheduled reports, custom dashboards, granular permissions, the natural next question is: what else can this replace? SQL Reporting Services. Microsoft Power BI You might be paying a subscription for a powerful analytics tool when your AI agent can likely build something fully customized, with no technical skills required, for a fraction of the effort.
And this is just the tip of the iceberg. Anything Claude can read, it can analyze, and turn into a system that pulls that data in real time. If you’ve ever wished you had instant answers to questions buried in your own business data, that capability exists right now. You don’t need to be an AI expert, hold a certification, or have years of experience. Plain language and a conversation with Claude is all it takes.
Brian Largent
Father to five, husband to one, founder, CEO, and all around swell fella (or so I'm told)
