FAQ: Responsible Use of Jett Generative AI
What is Jett?
Jett is an AI-powered advisor in Control-M SaaS that enables you to gain instant knowledge of your application and data workflows. Ask Jett workflow-related questions in simple language and get clear graphical and textual responses.
How does Jett ensure accuracy and reliability?
Although Jett leverages available data, follows AI best practices, and incorporates user feedback to reduce errors and improve recommendations, AI-generated insights are recommendations and not absolute solutions. You must review and validate outputs before implementation.
Is there any additional cloud data exposure if I use Jett?
No. Since you already use Control-M SaaS, all data is protected. In addition, the prompts and the retrieved data in the responses are not used to train the language models, which ensures data privacy.
What are the intellectual property (IP) considerations if I use AI-generated content?
Jett uses AI models that might be trained on publicly available code and documentation. Use caution and follow your organizational IP policies. Validate Jett content before you use it.
Does Jett generate code?
Jett does not generate code that is implemented in the product.
Jett uses language models to generate data queries to retrieve requested information. The system verifies that the generated data query complies with organizational and product security constraints. You must incorporate human validation to ensure result accuracy and relevance.
How does Jett summarize the response?
After Jett retrieves the data, it uses language models (LLMs) to suggest the top result highlights. BMC encourages you to review these highlights and consider other perspectives, as there might be additional important points that you must consider.
Can Jett introduce security risks?
Amazon Bedrock Guardrails are implemented to provide a foundational layer of protection for GenAI. They ensure compliance with security best practices to prevent unauthorized actions within the cloud environment. Control-M uses AWS managed services, such as Amazon Bedrock, which is the primary service for Jett.
Jett follows strict security policies and protocols wherein sensitive system data, credentials and other confidential information are not exposed to unauthorized users. Control-M does not store, process or transmit any Personally Identifiable Information (PII).
The solution was tested for common security vulnerabilities with various criteria, such as the OWASP Top 10 for Large Language Models. The BMC Product Security Group conducted a thorough penetration test of the solution. Role-based access control ensures that only authorized users can query and access data relevant to their roles.
Customer personal data, in Control-M SaaS, is subject to the BMC Data Processing Agreement (DPA) and processed according to EU and UK highest data privacy standards, as per the BMC Data Privacy Binding Corporate Rules (BCR). BMC is compliant with security best practices managed by AWS.
How does Jett prevent bias and ensure fairness?
Although Jett generates responses that are designed as neutral and unbiased to avoid discriminatory or misleading language, regular evaluations help detect and mitigate inaccuracies or biases in AI outputs.
Who owns the AI-generated output of Jett?
Users retain full ownership of their work. BMC does not claim any rights over Jett assistant-generated suggestions or responses. You should review generated responses to ensure compliance with licensing and attribution requirements.
Does Jett support compliance with industry regulations?
BMC is committed to responsible AI practices under the European Artificial Intelligence Act and has signed the EU AI Pact Pledge. For more details, see the BMC Responsible AI Statement.
By inheritance, Jett is a feature of Control-M SaaS and complies with the same industry regulations.
Is Jett intended to replace human experts?
No. Jett is designed to assist in human decision-making, not replace it. It enhances insights and recommendations but does not substitute human expertise. You should apply your own judgment, validation, and security best practices when leveraging AI-generated insights.
Can any Control-M user use Jett or are roles and permissions applied?
You do not need to assign special role to utilize Jett. After Jett is activated, it is accessible to all users. Jett responds based on individual user privileges to ensure that the reader only sees the information that they are authorized to view.
How can I provide feedback to improve AI-generated responses?
Jett features an adaptive feedback loop that allows you to report inaccurate, unhelpful or biased AI-generated responses. The feedback helps to continuously optimize the AI models, and enhance their accuracy, contextual relevance, and alignment with user requirements. Human-in-the-loop feedback enables the AI-generated responses to evolve dynamically and eventually improve quality and trustworthiness.
Does Jett work in a Unified View environment?
Yes. Jett works with Control-M SaaS, which also powers the Unified View to manage SaaS and Self-hosted workloads.
How can I have Jett enabled on my Control-M SaaS tenant?
You must first ensure that Jett aligns with your organizational AI policy. Jett offers a daily limit of 50 queries per tenant.
Contact your BMC representative to activate Jett at no additional cost.