Some large-scale invoice-related tasks were overwhelming our existing processes and attempts to handle them manually. We are in the data center industry and dealing with all these technology companies, but many processes in the real estate business are a bit more outdated. Many of our people are comfortable working in Excel, and some teams are highly siloed. One significant challenge I've faced as a UiPath developer is acting as an evangelist within the company. We want to demonstrate the platform's capabilities and get buy-in from these different teams across the enterprise to raise the level of what we're trying to do.
UiPath has been helpful with that first step of getting the information off the invoice. I've been learning and expanding my skill set on the workflow side. Many of our automations have a workflow with a human in the loop doing manual review. I look forward to automating between different departments, and that's one thing I want to develop at this conference.
We used to bring on seasonal contractors during peak seasons, and now we no longer need to do that because the existing staff have more capability. That was a big thing when we started. Our accounting team was so busy for the two weeks before and the one week after closing. There was a tiny window when they could engage with any process improvement or look ahead at what we could change because they were so busy keeping up with how things work. We've freed up these people who are intimately familiar with our business and give them more time to apply that knowledge instead of filling out forms.
We were processing about 2,000 invoices a month when I started. Now we're up to about three thousand. It took about 15 minutes per invoice to process because there were so many different elements. Working with Yardi is challenging. There was a big issue with getting these into Yardi and uploading them in batches. If one invoice in the batch failed, it would kick them all out. We spent all this extra effort troubleshooting and doing all of this.
Now, the bot can execute this work and upload them individually. If there's a single error, it can be isolated and kicked out as an exception. Someone can manually review it, and the bot can keep putting the rest of the invoices into the system. We've also had a great ROI on the monthly reporting. We generated reports from around 50 sites every week and then distributed them to a long list of different people on different projects. It's straightforward to do and only requires 10 clicks for each report, but it saves massive amounts of time for people. Now, all I need to do is maintain a list of who should get the emails and what projects need reports to run.
I love developing automations. I often directly help people by improving the part of their job that is time-consuming and dull. In addition to saving time, we reduce errors caused by manually typing things in. I've demonstrated that in different departments at our company.
We won an award for our ESG efforts. I developed an automation to help us report our ESG metrics because all of these customers want reports to give their shareholders about green initiatives. We wanted to take all the data on energy reductions in data centers and distribute it to each of our customers. We had a very complex template that we wanted to iterate on until we delivered the report. I developed a bot that could generate the source data and template of these files for our customers by data center and aggregate them.
Before I joined the company, they had no automation solution. They tried to do this with mail merge, which struggled because of the variability across our sites. We wanted to achieve greater complexity and offer this table of information when it's available or update it when the final file we're generating varies significantly. We wanted to be able to convert a Word doc into PDF format and aggregate all of those different PDFs at the site level and aggregate those per customer.