How we used ChatGPT to streamline our billing and save hours!

Since our launch 18 months ago, Digiphile has experienced remarkable growth. Our team continues to expand, we now serve over 80 clients, and we collaborate with local counsels globally.

Due to our rapid and continued expansion, the need to have a streamlined billing solution became increasingly essential. The solution we found, thanks to ChatGPT, helped us to automate the most complex part of our billing process (client e-billing using LEDES files) and knocked an entire day off our billing cycle each month. 

This blog explains the challenges we faced, how we found a solution using ChatGPT and how the power of AI, when used appropriately, can increase billing efficiency.

The Challenge!

As we onboarded more clients, monthly billing became increasingly complex and time-consuming. Those that used e-billing systems employed a variety of different systems, and there were even multiple configurations by clients that used the same system. This meant LEDES files generated by our time recording platform had to be individually edited to meet the specific requirements and format of each client’s e-billing system.

If the requirements were not met, the client e-billing system would raise an error and occasionally this would take an hour or two of manual investigation to resolve as the cause of the error was not always clear. Across multiple clients, this could result in days of extra work. As a result of the above, billing grew from a few hours each month to three days and looked likely to increase even further.

The Solution

Since we specialise in AI legal advice, we thought why not turn to AI for help?! Using our enterprise licence for ChatGPT, I asked the AI whether the edits we need to make to LEDES files could be automated.

To be clear, no client LEDES files or any client data were entered into ChatGPT. Instead, I simply explained to ChatGPT the types of edits we needed to make, which included things like automatically populating fields and performing simple tax calculations. I was somewhat shocked when, within a few seconds, ChatGPT provided guidance on automation scripts, along with a sample Python script to do this. It had never crossed my mind that ChatGPT could code!

Since we're not developers and do not use Python, we asked ChatGPT if something similar could be done using macros in Excel. “Yes, you can use an Excel macro (written in VBA)”, the AI replied, and it proceeded to code a simple macro for us using dummy data in the code as an example. Somewhat excited, I then adapted this macro locally on my own device as appropriate for each client.

This meant, after a simple discussion with the AI, it had produced code that could be easily customised to meet all our requirements. We had hit the holy grail of automating edits to our LEDES files, but there was still one giant problem. I had no idea how to run the macro in Excel since I had never used macros before!

If ChatGPT can code, I thought, I’m sure it can provide a list of concise bullet-point instructions that teach me, with zero knowledge of macros in Excel, how to implement it. Indeed, the instructions were so clear that it was impossible to get wrong, despite my best efforts.

After customising the ChatGPT code and replacing the dummy data, we now have a list of macros that meet each client’s e-billing system and configuration requirements. Implementing the macros is as easy as opening the LEDES file in Excel, clicking Macros on the Developer tab and selecting the relevant macro. The macro itself can be run in a matter of seconds and then I spend a few minutes reviewing the output for quality control purposes to make sure everything is as expected.

Increased efficiency with AI

Since this, I now use AI on a daily basis at work to help with productivity issues. This helps me to do far more work in the same amount of time. In accordance with our internal policy, I never enter any client information and instead mostly use AI as an education tool (such as how do I do x, why is y the answer, is there a better way of doing z) and look for clear, factual, evidence-based answers.

It still amazes me how one question led to the complete transformation of how we approach billing, saving a whole working day every month and I wonder what other questions I haven’t asked that could save even more time.