AI Agents: Things You Should Know!

AI is going beyond the prompts. And they are now allowing you to do things in the automation realm too. Like you have a flow of work from A to B. And in this flow they want to do things with the help of an AI.

In this approach they have learned how to do the step wise work. And they can fill in the blanks to understand how to. And not expect the humans to guide them.

For example, say you have pdf to download from xyz site and then read it’s 10 pages. So the thing that would be working here would be getting bot to go to website then find the pdf link and then getting it to click for download and then reading it. So this would be of many steps.

But the prompt engineering aka writing the prompt would only involve the point steps which does the task. And not the steps in between. AI agents are to be trained for steps in between. That is the form of intelligence they would learn.

Google Colab recently did the hackathon that worked on the AI agents part.

Here the challenge was to get the AI to do certain steps. And then get them to solve the issues as well. Make sure to do the AI steps for the work. This can also be done as a testing steps that people do with the manual testing or automation.

AI Agents can do variety of things but as the models are being built and the AI improves it’s decision making and the stepwise calculation of the action. Things would get sorted as well. Which is what is expected here.

AI agents are designed for those automation scenarios. Sooner or later they would do a lot of our local tasks too. It is just a matter of time!