How to create and automate with AI with the chat interface and nothing else

AI in its current state is a wonderful tool for automation, but like with any other new tool, you have to know how to use it safely and efficiently.

You can use AI to automate your everyday tasks, with no coding, no deep knowledge of tech magic behind, with a simple access to one of the main LLMs through their chat interfaces. However, using it without basic understanding of its limitations might lead to repeating mistakes many of us have already made, and wasting your time unnecessarily.

There are certain AI limitations you have to be aware of:

• context window limitation,

• bias to comply and flatter,

• lack of strict rules to follow.

Let’s dive in, explore what it means, and how to overcome the challenges it brings. Today, we will cover:

Context window limitation

There is only a certain amount of tokens the LLM can process at once. If your conversation, including the first prompt, is getting too long, the AI will automatically shorten the oldest parts. That’s an efficient way of handling it, but it can lead to an increasing amount of mistakes and lower accuracy. Important bits of information can get lost in the process, the AI “forgets” instructions from the beginning of the conversation and you have to remind it - if you realise that happens in the first place.

How to handle it?

If you’re discussing something with AI and you reach a certain point from which you want to handle a couple of similar tasks, start a tree conversation with multiple branches, every task having its own branch, with the shortest possible way to reach the important information at the beginning.

Example: You prompted chatGPT to use your voice and create emails that look just like yours. It took some time, you had to feed it with examples etc. Now, it delivered the first usable email - let’s call this the “point 0”. From now on, you will want to use it for handling your correspondence, telling it “John told me he can’t make it to the office today, let him know I am pissed, but I understand it’s necessary” or “respond to this email, saying politely NO”. Every task like this should be right below the “point 0”, not lower. Therefore, whenever you want to do this, don’t send a new message in the current thread, change the message right below the “point 0” to your current task and resend it. The AI will create a new branch of your conversation, disregarding those 13 emails you have already created, but having the good example fresh in the thread, remembering all the setting etc.

We will cover the rest of the topics in the upcoming articles. To make sure you don’t miss it, you can subscribe for receiving our newsletter and receive a free PDF with advice on AI automation here.