TFF #010 - Artificial Intelligence vs Robotic Process Automation
Read Time - 4 Mins
This week’s newsletter is written in Notion, which I use as my notepad, and my task manager.
I use Notion AI (which uses ChatGPT) for research, inspiration and to help structure my thoughts.
So, even though I have an AI assistant, the words you see today are my own.
This week we’re looking at the differences between Artificial Intelligence (AI) and Robotic Process Automation (RPA)
Is the development of algorithms and systems that can perform tasks that would normally require your brain.
These include: recognizing speech, solving problems, and making predictions.
Machine learning, deep Learning and Natural Language Processing are AI.
I think of AI as a digital brain.
Is the use of software bots to automate routine tasks that would normally require your hands.
These include tasks like clicking buttons, typing on a keyboard, and running Excel macros.
Repetitive Data Entry, Data Processing and Data Integrations fall under RPA.
I think of RPA as a digital set of hands.
What is ChatGPT?
ChatGPT uses a combination of deep learning, and natural language processing.
It is a text based bot that can understand questions and surface answers based on it’s training.
Therefore ChatGPT falls into the category of AI.
But, it is also currently being built into other tools where it’s also moving into RPA territory.
For example, you could connect it to something like Xero and ask it to run monthly journals.
Or you could ask it to find and correct an invoice.
The latest podcast ‘Creating a Junior Accountant using AI (ChatGPT)’ with Chris Bell covers it in detail.
If you missed it, you can find it here.
So what should you use & when?
1/ What are your goals?
AI is built to TRANSFORM businesses and industries.
RPA is built OPTIMISE existing processes.
So if you’re already automated and looking to take your business to the next level with trend analysis and insights, AI can help you.
If you’re still stuck with a load of repetitive admin, start with RPA.
2/ How difficult’s the task?
If you’ve got a task that’s well defined and can easily be explained using rules like “Copy data from point A to point B” then RPA could work well.
If you need to simulate a brain for deeper analysis and pattern recognition like “What are your cashflow predictions for the next 6 months” then AI is the way to go.
2/ What’s the volume of data?
AI performs at it’s best when it’s got a lot of data to work with as predictions become more accurate, the higher the volume of information.
So if you’re looking to make decisions from a lot of data, then great.
RPA works better for smaller, more targeted datasets.
So if your aim is processing smaller amounts of data and keeping it accurate, RPA may help.
4/ How flexible does it need to be?
AI has the ability to adapt, learn, and handle changing datasets.
RPA is limited by the rules and scripts it’s been programmed to carry out.
So if the task is more fluid and requires any form of change, then AI is the way to go.
If the required task is highly structured and predictable, then use RPA.
Use Cases for AI & RPA
Predictive Analytics - E.g Planning & forecasting - AI can analyse past data and use it to suggest future trends
Data Accuracy & Approvals - E.g Month-end close - AI can analyse trends and flag when transactions are inaccurate
Auto-categorisation - E.g Expenses - AI can auto-categorise transactions and also flag suspicious ones
Writing documentation - E.g Job Descriptions - AI (In this case ChatGPT) can auto-write a document “Write me a job description for a Financial Controller” that you can then tweak
Import / Export - E.g AR & AP - RPA can automatically send payment files to your bank.
Reminders & Notifications - E.g Dunning - RPA can automatically send reminders to overdue customers.
Reporting - E.g Management Accounts - RPA can aggregate data from multiple sources and run run scripts to format the data correctly.
Some use Both
OCR (Optical Character Recognition) - Will use AI to decode information on an invoice, and then RPA to enter the invoice.
Auto Bank Recs - Will use AI to identify similar transactions, and then RPA to match them in your system.
1/ AI is great if you're already optimised and are looking to further transform, using technologies that function like a human brain.
2/ RPA is great if you're not optimised, and are looking to automate repetitive tasks, using technologies that function like a virtual set of hands.
3/ Deciding on which one to use, depends on the complexity of your task, the volume of your data, and the flexibility of processing you require.
4/ Chat GPT is an emerging AI technology that when built into other applications crosses into RPA territory. But it can also be used for personal / admin tasks too.
That’s it for this week.
If AI & RPA are a priority for you right now.
I’ve 2 things that might help:
If you’re looking for guidance - A free 30 min general chat with me about what tech could help you. You can book a time here.
If you’re looking to upskill on ChatGPT - My friend Nicolas Boucher has just released a 92 page guide on ChatGPT with 22 real uses cases for finance pros + step by step instructions. You can get your copy here.