Generative Recorder FAQs

Review the FAQs to learn more about Generative Recorder.

When will Generative Recorder be available?
Generative Recorder will be available for all licensed cloud customers from Automation 360 v.32 onwards.
How do I opt for the Generative Recorder feature?
Generative Recorder is available with the Automator AI suite.
Can I use Generative Recorder in my On-Prem Control Room?
Generative Recorder is currently available only to our cloud customers. Generative Recorder support for On-prem Control Room is in roadmap.
Do I need to purchase any licenses from Generative AI providers or perform any LLM configuration, to use Generative Recorder?

No, Generative Recorder takes care of Generative AI integration. There is no need to purchase any extra licenses or perform any additional configuration to enable Generative AI-based fallback.

If the company policy does not allow using Generative AI, can I still use Generative Recorder?
Yes, you can use Recorder native fallback to get enhanced automation resiliency. Native fallback is not based on Generative AI.
What is the difference between Recorder native fallback and Generative AI-based fallback?
  • Recorder native fallback: Relies on additional UI element details gathered during automation design. It employs Recorder's built-in algorithm and this extra data to automate UI elements if a fallback occurs.
  • Generative AI-based fallback: Utilizes the current state of live applications, excluding sensitive business data. It employs large language models to identify unique characteristics of the target UI element for automation in case of a fallback.
Can I try Generative Recorder before buying?
Strategic customers can try out Generative Recorder through our cloud hosted preview environment.
Can I selectively enable Generative Recorder per automation?
Yes, Generative Recorder can be enabled on a per-automation basis according to your needs and preferences.
Is there a circumstance where Generative Recorder might not work?
If the application has undergone such a drastic change that the AI cannot find an equivalent path to what it has seen before, it will fail just like the normal Recorder does. In such cases, we recommend that you redesign that part of automation.
Which Generative AI provider is used by Generative Recorder?
Generative Recorder is designed in a way that the choice of underlying Large Language Model is completely transparent. We currently use Microsoft Azure OpenAI. Note that there is no direct integration between your data center and Azure OpenAI as all the outbound communication happens through a dedicated service hosted within the Automation Anywhere Cloud.
Can Generative Recorder integrate with the Generative AI providers that are approved in my workplace, for example, from Google, Amazon, Anthropic?
We are currently evaluating this capability for our roadmap.
How do you ensure the security of application business data for Generative AI-based fallback?
For Generative AI-based fallback, we use only the application source code data that is specific to the currently automated UI element. We do not retain, transmit, or use any business data, for example, personally identifiable information from the automated business application. Any potential business data that is a part of application source code itself is used only after anonymizing it. Additionally, any data entered or extracted by the automation until the fallback occurrence is also excluded from our processing. We ensure that no business data is routed to LLMs.
Does Generative Recorder open any direct communication between my company's datacenter and public Generative AI providers?
There is no direct integration between the datacenter and public Generative AI providers. Instead, any fallback requests to Generative AI are channeled through a dedicated service hosted within the Automation Anywhere Cloud. This provides better control over the data exchanged with Generative AI providers. Consequently, this enhances data privacy and security, as any potential sensitive information is not directly exposed to GenAI providers reducing the risk of data breaches or unauthorized access.
Generative AI has a hallucination problem. How can a I trust it to provide a reliable fallback for our automations?
We use a series of carefully designed prompts that limit the scope of what paths an automation can take. This prevents hallucinations from coming back from Generative AI altogether. When in doubt, the automation will not guess.
Does I get a record of instances where Generative Recorder prevented automation failures?
Generative Recorder provides complete traceability to authorized users for all the instances where it helped prevent automation failures. This includes notifications in system alerts with a link to the details of the specific automation. Audit Log entries also document relevant fallback details, such as the user's identity, device information, automation name and the line number where fallback occurred. This comprehensive record-keeping facilitates the monitoring of instances where Generative Recorder played a role in influencing automation decisions.
Will you be caching Generative AI responses to save cost for automating the same UI element in the next run?

Yes, we will be caching Generative AI responses for successful fallback events. This will save on Generative AI API calls and also reduce automation execution time. This will be supported from Automation 360 v.33 onwards.

Can Generative Recorder auto update the automations, per Generative AI recommendations?
Generative Recorder enables auto update of automations with a human-in-the-loop. Citizen Developers and Professional Developers get the list of recommendations to auto update the automation to make those more resilient. Automations are updated after recommendations are accepted.
Do I need to open any ports to use Generative Recorder?

No. If you can access Automation Anywhere Control Room over cloud, you get access to Generative Recorder too.