1. What is process data?
Process data are user actions taken on the desktop or web, transformed into machine learnable data.
2. What kind of training is needed to implement officeAutomata Automation?
No user training is required to for officeAutomata AI to be trained as the AI runs in the background of user systems and records every user action. For advanced AI actions like OCR and web scraping, walkthroughs are provided in the main dashboard. For management, an interactive walkthrough of the dashboard is provided.
3. Who is responsible for the quality of automations produced by the AI?
Training time of the AI can vary and directly impacts the quality of automations that the AI produces. More training time will result in higher quality automations. OfficeAutomata is responsible ensuring that AI is being trained accurately and users are responsible for training the AI to produce the automations that they want.
4. How does an automation start doing a sequence of tasks?
There are two ways to start an automation, a predefined trigger or manually. Triggers can be almost anything; a user action, sequence of events, email from a client, a directory change, a file changed etc. The trigger you select for AI automations will depend on the type of work you are training the AI on.
5. Has officeAutomation created automations for itself and how well do they work?
OfficeAutomata’s staff uses our AI for data entry, web scraping, expense tracking and voice commands. Most of our work has to do with backend coding, but where we can use our own software to make tasks easier, we do and it makes a world of difference to automate time consuming tasks.
6. What if an automation starts deleting work files and information because that is what you were doing previously?
We suggest our customers take the time to review automations before putting them in place. This sort of quality assurance will help initial automation accuracy and give more information to the AI to learn from good examples.
7. Is there some way to prompt an automation to be produced if you’ve done everything to give the AI process data?
We call this manual automation and we provide tools to do this. Employees have the capabilities to record events manually and then save those events on the event logger. This is a little more time consuming but the results are the same.
8. What does Automation do with ambiguous process data? And will the Automation realize that it is ambiguous?
There are confidence levels associated with each automation produced. The more data and repetition for a single task that is given to the AI will increase the confidence levels for the automation produced for that task. This helps make ambiguous data less ambiguous to the AI.
9. If you only want part of the steps in an automation can you change it easily? And how would you go about changing a step in an automation.
It can be very simple to change the automations that are produced. Events can be taken out, deleted or replaced. This is a key part of managing automations.
10. If an automation runs but it messes a bunch of stuff up, is there any way to reset?
Depending on the automation, it is possible to reset data to their original state. The AI can even be training to do more complicate resets.
11. How much process data does the AI need to create an automation?
The AI can learn with even a few days of training, but more training time will increase the confidence of building automations for the AI and an employees can be more confident in the accuracy and quality of the automation produced.
12. What if the AI isn’t learning enough before it creates an automation and it keeps producing bad automations based off of too little data?
You should just have to wait, because the AI continues learning. It is possible to make notes to the AI to help refine what automation you want and what quality you want.
13. Would viruses affect the AI and automations?
No, viruses would not effect the AI or automations, but disrupt AI learning from users if they were fighting a virus.
14. What if you are working on two computers at the same time on the same task? Would some process data not be logged?
This should be a non issue if the user is on the same profile, because automations and AI learning are associated with user accounts.
15. What if you are working on one computer but multiple tasks at the same time? Would the process data be confused?
Time, users and repetition will all affect multitasking. With enough users and training, the AI will be able to accurately build automations for the seperate tasks.
16. Is the deep learning AI only for creating automations or is it or can it be used for other parts of officeAutomata’s services?
For the moment, our AI just works on producing automations. There are lots of other possible uses for our AI and in the future we will be producing other types of AI based on customer experience and feedback.
17. Are the features and services used by the AI only for the AI or can individual features be used? Like pdf extraction and OCR?
Our customers will have to use those tools in order to teach the AI and produce process data for the AI. For our customers this is a win-win, not only do they get an AI that produces automations but they get excellent computer tools such as OCR, text analysis, image recognition and lots more.
18. If there was a system crash or something was down for the day for maintenance what would be able to work?
Because of the nature of some of our enterprise clients and their work for the healthcare industry and healthcare privacy laws, enterprises run our software off of their own servers. For our Team clients our software runs on the Cloud and individual clients are able to download our lite version of officeAutomata to their PC.