Why We Build: M&A and Bots

Welcome to the first inaugural post of my “Why We Build” series, where we go into the details and business problems our clients are facing and why they’re engaging Pato Solutions for help solving them. For context, I am a professional software engineer with a decade of experience working in the Fintech and E-commerce industries, where I built web, data analytics, and data science applications. I left my corporate job in 2023 to bet on myself and I currently offer my services to those with challenging problems that need help. Let’s get into it.

It was mid January when I reached out to a client inquiring about getting tax forms when I first caught wind of the current pain point they were facing. They had just acquired a new business, expanding their locations from 2 to 3, and were mercilessly working trying to integrate their business processes and systems. Unfortunately for them, they did not have much of any assistance in migrating data from their current software and ERP system providers. They were looking down the barrel of an expensive and logistical nightmare trying to reconcile client, parts, and sales history lists. I offered to use my software skills to help.

I had been building operational reporting off of their IBM DB2 database and ERP application for several months now, so I knew in general what we were working with. My first step was to try to see if we could insert data into their systems using the same database connection configuration we had been using to read data into Power BI (the analytics system powering their reporting). Unfortunately, we were unable to find a configuration that allowed for write access directly to the tables (which I expected), and I could also not find any stored procedures that I could leverage so write that data in an “approved” manner (which was unexpected). The easy solution was out the door.

Next, we entertained more creative solutions. I studied the guts of the website and devised a plan wherein we mimicked user interactions using software – a data entry bot. I called my client back to tell him of this viable path forwards. “I need to know in two hours if this can get done, so I can send people home”, he remarked. I rolled up my sleeves and, relying on the years of web application software experience, programmed a working prototype in the next couple of hours. I gave him a call back letting him now it is doable, and I’ll have his customer data loaded by next morning.

Over the next couple of weeks, I worked with this client to load their precious data enabling new employee onboarding and giving existing employees a comprehensive view of the businesses. In order to accomplish this, I designed a system where I could queue up a specific task (e.g. “Add new customer”, “Update inventory locations”), and later assign workers to run and execute those tasks until there was nothing left to process in the system. At peak times we have 8 simultaneous bots commandeering their own web browser page and adding about 50 records / minute. This is easily 10 times faster as a corresponding human with a well optimized workflow, and the bots don’t need to take breaks. It wasn't without its headaches as any bad input, or unexpected application validation state, caused the bot execution to come to a screeching halt, but we mitigated it by sending those types of records to a "dead-letter" queue, to be evaluated and reprocessed later.

I needed a disciplined approach to minimize mistakes, and it was invaluable being able to pull data from their database and cross reference it to the configuration files. It was incredibly rewarding to run an audit script that I had just drafted up, send my client excel files of the different categories of data issues, get feedback and direction on how to handle those problems, make a couple tweaks to the bot workflows, set them loose, and rerun the audit observing the problem count plunge towards 0.

If you are responsible for running a business, I would highly recommend you keep a friendly software professional like myself plugged into your business. With a breadth of experience from building data intensive web applications at startups to running data science experiments at one of the world’s largest companies, I can tackle a wide variety of problems and get you the results you need. As we like to say, “I wish I could do this better... Reach out, you just may figure out, we can”.