Table of contents
Working on Square AI has been one of the most rewarding projects of my career. After a decade as a startup founder and early operator, building Square AI feels like being back in the saddle — helping new technology cross the line from possibility to reality for everyday business owners.
How it started
Square AI began the way many of us first engaged with tools like ChatGPT: experimenting, playing and asking “what if?” During a company-wide hack week, employees from across Square came together to prototype ideas for a business “copilot” — a tool that could answer questions, analyse data, and even take actions on your behalf.
The early demos generated huge excitement internally. But we quickly learned a critical lesson: demos are easy, building products is hard. Large language models (LLMs) make it simple to build impressive prototypes, but real products need to handle the messy, unpredictable questions business owners actually want to ask. Conversational AI also doesn’t have an interface to teach you what you can or can’t do — you just have to get it right, or explain why you can’t.
Finding our focus
Our small team gained the mandate to push forward on Square AI soon after our demo. We partnered with a dedicated group of business owners we call Square Champions. These Champions gave us their time to test buggy prototypes, brainstorm pain points, and imagine what an “AI hire” might do for their business.
One theme from their feedback was searingly clear: business owners have more data than ever but not enough time or clarity to use it. They wanted help finding patterns, problems, and most of all opportunities.
Up until now, business operators have had to just sit there and wonder. We asked – what if they didn’t have to?
Previously, I’d pull reports from Square, wrangle them in a spreadsheet, and manually match them with deputy labour costs. It would take 30–60 minutes easily to get a usable heatmap. Square AI gets me there in five minutes, so we can keep staffing tight and stay on top of labour costs week to week.”
Gabriel Larraz → Co-founder, Broken Eggs
Zeroing in on insights
Armed with feedback like this, we focused Square AI on a single problem: helping businesses answer questions about their data. To start small, we focused just on sales data, and narrowing scope meant we could move extremely fast. In only a few weeks, we built a proof of concept that could take a business operator’s question, translate it into a request against internal APIs, and return the answer.
We hosted live “playtest” sessions where we took our prototype to real businesses and prompted them to ask real questions. These sessions exposed challenges right away: What defines a “week”? Do “daily sales” mean averages or day-by-day totals? How should time zones be handled? We were also able to observe how business operators actually wanted to interact with AI, for example how one question often built on top of another to take you in a different direction than you initially thought. The engineering team participated directly in these sessions, and we addressed feedback within days.
We also got a chance to see what created delight right away. Business operators loved getting clear, direct answers to a specific question vs. having to work backwards from a chart or big table for what they wanted. They loved being able to just say “Christmas” or “August Bank Holiday” rather than remembering the date. When early testers started asking if we could take screenshots and send them the data, we knew we were on to something.
We rolled out Square AI to two of our venue managers, and they’ve become much more proactive with weekly reporting. It’s faster, easier and it’s helped us make more precise inventory decisions based on accurate sales data.”
Jayke Mangion → Owner, Brickwood Coffee & Bread
From prototypes to beta
By mid-2024, we had an alpha version testers could use inside their Dashboard. Square businesses could type a question in plain English into Dashboard’s search bar, and get a clear response with their own data. We enabled this for the same Champions who participated in our early testing sessions, and we saw them coming back. They started sharing their answers with us and with each other in the Square Community, and telling us how the same question would have taken them hours of time, or taught them something new. While early, we had a clear signal there was a path to product-market fit.
We launched a closed beta in August, inviting more Square businesses in small waves. Every onboarding session doubled as user research, and each new request shaped our roadmap. Our early testers tried asking about labour costs, customer behaviour, transaction details, or simply questions about Square, and we added each functionality within months. Most importantly, we realised business owners didn’t want to search for data. They wanted to have a conversation about it. So we rebuilt Square AI as a chat experience.
Staying close as we scale
As adoption grew, we scaled the beta programme while preserving weekly conversations with active users. These calls continued to surface needs we couldn’t have spotted in data logs. Businesses asked to save answers, ask questions on-the-go, or incorporate external data like local events or reviews. Those requests directly shaped features in latest updates, from conversation history to web data support.
In May 2025, we made Square AI broadly available as an open beta, and today, Square businesses ask thousands of questions every day. Still, we make it a priority to bring them into the conversation early, and every feature you see has been moulded by feedback from businesses just like you from the very beginning.
Business owners have said things like “I could see myself moving over to the Dashboard App for most of my Square AI queries” or “local weather and events can make or break our day. It’s so nice to be able to ask about them right in Square AI.” We can’t wait to hear what you have to say.
Why it matters for your business
Local businesses face the same complexity as large enterprises, but with fewer resources. Owners juggle marketing, bookkeeping, customer service, and operations. Square AI is our small, early, but ambitious start to being your thought partner — part analyst, part CFO, part operations manager. It’s meant to amplify what makes business owners so special — their curiosity and drive — and help you translate all that wondering into impact.
In addition to just making it faster to get data you already review today, Square AI makes it easier to dig into details you might have relied on your gut for before. Instead of having a general idea of Thursdays, you can now drill into every hour across all Thursdays, at a particular location, in a particular month, or for a particular metric.
Where we’re going
Square AI is already saving Square businesses hours on reporting and unlocking insights they couldn’t reach before. But we’re just getting started. The next frontier is proactivity.
A good COO answers your questions quickly and accurately. A great COO brings you insights you didn’t know you needed. That’s where we see Square AI heading — not just waiting for your prompt, but surfacing opportunities, warnings, and ideas on its own.
We’ll keep building this hand in hand with business owners, in coffee shops, restaurants, retail stores, and beyond. Not as a tool, but as a partner. Learn how here.
