WeOS

Empowering Square users with self-service data analytics

Comment 1

Inspiration 🤔

Running a business is hard...I KNOW!!!

As a serial entrepreneur, I know running a business is hard. It's not a 9 to 5 job. You think about it almost all the time and sometimes there are things you simply want to know on the go. You may be at your child's chess tournament and a question about an order popped into your head or you want to noodle something about your sales for the week so far. I've been there!

You can certainly go into your Square app, navigate to the orders section, search for the order in question and find the answer you're looking for. But what if you could just simply ask "What was the average sale per customer last week?"

What it does 🔥

WeOS lets you chat with your Square data in minutes

No more wasting time finding the right screen or fiddling with filters to get your desired answers. If you want to know something...JUST ASK! Our simple chat interface allows users to get direct answers fast whether they're on a laptop, tablet or mobile phone.

For example you can ask: "How many new orders did I get last week?" "What was my best selling product last month?" "How many new customers did we get this week?"

How we built it 🪄

WeOS was built by leveraging what Google AI is good at and integrating it with useful Square data

Square APIs Used for Data

  • Catalog
  • Employees
  • Customers
  • Inventory
  • Orders
  • Payments
  • Refunds
  • Disputes

Google Vertex AI as the LLM for query interpretation, query generation and response

Challenges we ran into 💪

Google AI The non-deterministic characteristic of Google AI is a double-edged sword. It does a superb job at interpreting queries, however, its responses can be inconsistent if guardrails are not in place. So to ensure accuracy in responses, we've found a clever way to leverage its amazing ability to interpret user queries and with the right context and information, it can produce useful, consistent results.

The Square API is deeply nested and APIs are optimized for performance though they lead to having to make many calls to get the full information.

Accomplishments that we're proud of 🙌

Taming the AI beast (to some extent)

  • Getting the AI to return consistent results
  • Getting the AI to learn user-specific workflows

Using more of the Square API

  • Integrating with a broad range of Square APIs

What we learned 🤓

  • The Square user community is very active
  • Square APIs offers a breath of functionality
  • Google AI is developer-friendly

What's next for WeOS 🤩

We think this would be an excellent marketplace app Before we do, we want to do a few things

  • Develop a scalable way to provide user support
  • Expand the Square integration so that business owners can ask anything their data in Square can answer.
Share this project:

Updates