Q&A Part 3: Ensuring data security with Tray Merlin AI

Tray Merlin AI
Merlin
Artificial Intelligence
AI
Conversational automation
Natural language automation
Ensuring data security with Tray Merlin AI
Ali Russell

Ali Russell

Everything you wanted to know about Merlin, the new natural language automation capability in the Tray platform.

We recently held the Tray Megacast: An in-depth look at Merlin AI, where our CEO, Rich Waldron, and I unveiled Tray Merlin AI - our revolutionary natural language automation capability in the Tray platform. 

(If you missed the event, no worries! Watch it now on-demand). 

After seeing Merlin in action, many of you had interesting and insightful questions about how this new capability could enhance business processes and bring unprecedented efficiencies to your work. Since we didn’t have time to answer every question during the live event, we’re addressing them here, as well as expanding on some of the incredible capabilities Merlin has to offer.

In Part 1, we defined Merlin, discussed some of the possible use cases for conversational automation, and provided an update on Merlin’s availability. In Part 2, we discussed the many ways in which Merlin is a top-notch automation assistant by providing suggested code snippets, best steps, and more. In this part of the series, we’ll dive deeper into Merlin’s security and governance. Let’s get started!

Ensuring security and governance with Merlin

Q: Is the data or conversation with Merlin confined to the organization?

A: Yes, ensuring the confidentiality of your data is of utmost importance to us. With Merlin, all data and conversations are strictly contained within your organization. We have implemented robust security measures to maintain the privacy and integrity of your data throughout your interactions with Merlin. 

Q: What steps are taken to ensure that no data leakage occurs between clients? Are the source data sets used for the learning models entirely separate? 

A: We understand the significance of data security and the need to prevent any form of data leakage between clients. To ensure complete data segregation, Merlin and the Large Language Model (LLM) are not trained on any customer-specific data. We maintain a clear separation between customer data and the learning models used by Merlin. 

Customer data, other than what the user prompts, is not sent to the LLM, eliminating any potential for data leakage as customer-specific data remains isolated within your organization’s boundaries. We follow industry best practices and employ strong data governance methodologies to safeguard the confidentiality and privacy of your data.

By leveraging the Tray platform’s existing governance model and workspaces, we enforce strict access controls and provide granular permissions. These measures help maintain the separation of data and ensure that each client’s information is stored and processed independently. Our commitment to data security and privacy is an ongoing priority, and we continuously enhance our safeguards to protect your valuable data.  

Q: How does Merlin handle permissions when each person has different access levels to various tools?

A: Merlin operates within the existing governance framework of the Tray platform, which includes robust security and access management features. Just like with the other Tray platform functionalities, you can leverage workspaces, permissions, and roles to control who has access to Merlin and which authentications they can utilize. This ensures that users only have access to the tools and services they are authorized to interact with, maintaining data security and governance across your organization.

Error handling and resolution

Q: What kind of organic error handling or suggestions can Merlin produce? How does Merlin handle validation or errors when constructing a workflow? 

A: Merlin has built-in error handling mechanisms to help you address issues and refine your workflows. When encountering errors, Merlin will attempt to self-heal and suggest potential solutions within the chat interface. The user can then evolve or provide updated instructions to amend the tasks without having to rewrite the entire prompts. By leveraging its generative AI capabilities, Merlin aims to make the error resolution process more intuitive and efficient. 

Q: Can Merlin reconfigure connectors after analyzing a log output error?

A: Yes, this is a feature we have been developing.  It is not yet in production primarily because it involves passing the error message into the LLM, and we are avoiding sending potentially sensitive customer data into the LLM.  We are looking at ways we can obfuscate sensitive data or allow the customer to override this feature.

The roadmap for Merlin

Q: Are there plans for Merlin to support external API documentation for connecting to apps without native Tray.io connectors?

A: Merlin already supports the ability to send raw HTTP requests to APIs for which we don't have a connector. The main issue arises with OAuth-based APIs, as the user would be required to provide OAuth credentials beforehand, which currently involves setting up a custom service on the Tray platform.

Q: Are there any plans to add connectors for LLM ecosystem services/tools like Pinecone or other vectorDB providers?

A: While we currently provide LLM connectors like OpenAI, our roadmap includes expanding our ecosystem of connectors. We are continually evaluating customer feedback and industry trends to identify opportunities for new connectors. While we do not have any immediate plans to include Pinecone or other Vector databases, our connector builder tool and CDK enable you to build connectors for various services. Additionally, we have future plans to enhance support for developing AI applications directly on the Tray.io platform, which will involve features like chaining prompts and constructing AI agents.

Looking ahead

Merlin signifies the beginning of a new era of automation, giving anyone the power to interact with your iPaaS using natural language. Merlin’s ability to understand complex queries, suggest and generate code, build powerful workflows, and rectify errors makes it an invaluable asset for businesses looking to transform their processes.

While we covered a lot of ground at our Megacast and in this blog, this is just the beginning. As with all powerful tools, the true potential of Merlin lies in its application. We can’t wait to see how you’ll leverage Merlin to revolutionize your business operations. Watch the full demo of Merlin and stay tuned for more updates!

Subscribe to our blog

Privacy Policy