As the owner of a consultancy providing intelligent automated solutions to clients, it weighs heavily on my shoulders to maximize client ROI and ensure the solutions we create solve problems in the most efficient way—both for us as developers and the client as the end user. We write, on average, 500 lines of code per automation for our automated solutions and have even written as much as 1,500 lines of code for a single, more complex automation. However, with generative AI tools now at our disposal, why not work smarter instead of harder?
Many of my peers in the automation realm have yet to take the plunge with their first generative AI use case. And I completely understand the caution. But having become somewhat of a ChatGPT connoisseur, I’ve witnessed the smart, powerful, and just plain neat things this generative AI tool can accomplish. I want to share some of my insider tips and discoveries in hopes that they may put some of your hesitations at ease and inspire you to dip your toes into the generative AI pool. I promise the water’s great!
Intelligent Document Processing
For most of my clients, documents with unstructured data present the biggest challenge from an automation perspective. Before generative AI, those use cases with unstructured data would be pushed to the bottom of our pipeline because there was no concise solution. Since tapping into the power of ChatGPT, I’m happy to report I’ve seen wild success in addressing unstructured data with efficient and accurate automations.
For example, our clients do business with global vendors, which means they regularly receive documents with dates written in umpteen different formats. Traditionally, our developers would have to code an automated solution with ten or more individual “if” statements to parse this data. Even then, a few date formats would inevitably still be missed. With generative AI, on the other hand, we simply extract whatever date formats the client has received, run them through ChatGPT with a prompt to format the dates in a certain way, and pull the results back into the database. We’ve gone from 10-15 lines of code that wouldn’t even capture everything to only two lines of code with approximately 99% accuracy.
Easy as API
If you’ve ever developed automations with large ERPs like Salesforce, you often need a secret client ID and/or password, etc. But if you’re automating with ChatGPT, all you need is your API key, and that’s it. Easy as A-P-ie!
Create your own private GPT
This is a brand new development as of only a few months ago. If you have ever used Confluence or similar software to gather your data into one repository, you can now do this with ChatGPT. Once you sign up for the company-wide program, you can input all your data and ask it questions based on the data you’ve provided. Now that GPT has become specific to your organization.
Let’s say you have a policy document and you’ve put all your HR policies into one GPT. The questions or prompts will all be limited to that set of data and your automations can make requests from this specific GPT rather than a general one. This allows greater control over the security and competency of your data.
Creative Jump Start
I admittedly love playing with the photo creation capabilities of ChatGPT just for fun, but recently, I’ve applied this creative feature to benefit our clients. Working with graphic design firms who often design logos and branding for their own clients, we have been able to help them engineer prompts for ChatGPT to create very primitive logos and other graphic art concepts that they can use as a jumping-off point rather than starting from scratch. I foresee this having many applications in the marketing and creative arts world—not necessarily to create the final product, but to give a significant head start or boost of inspiration.
Ready to dive in?
I hope these tips and use cases I’ve shared for ChatGPT today have sparked ideas and confidence in how you can drive value with generative AI in your organization. In all reality, I’ve only scratched the surface of ChatGPT's capabilities, and I've only addressed one tool within the burgeoning family of generative AI products. With each passing week, I’m discovering more that can be accomplished and reimagining how we can transform business operations across functions and industries with generative AI added to our tech stack. So, are you ready to chart a new path with generative AI?