Mayvn AI
Frequently Asked Questions
Mayvn AI has been built by OFS with the express aim of safe, accurate and secure access to Generative AI.
In summary:
-
Privacy of customer data is built into the system architecture for Mayvn AI.
-
Mayvn AI features are operated exclusively in AWS environments which are:
-
Architected for compliance, and resilience, meeting SOC and NIST standards for physical, logical and architectural security.
-
Trusted by enterprises, governments, and SMBs worldwide.
-
Private to OFS and not shared with any 3rd parties.
-
-
You have complete control over the data held by OFS. Data is only provided to AI features when you interact with them, and nothing happens when you don’t specifically request it. If you want your data exported or deleted, just ask!
-
Mayvn AI is governed by existing OFS Terms of Use and Agreements.
-
No changes to these terms are proposed or required to start using Mayvn AI features.
-
-
If you don’t want to use Generative AI features, they can be turned off for you.
What is Mayvn AI?
01
Mayvn AI is a new set of features launched by OFS to offer you secure, embedded access to Generative AI within OFS Analytics. You can interact with Mayvn AI to ask for summaries and insights which are identified from the data already held by OFS about your manufacturing operations. OFS have built tools for you to converse with Mayvn AI, save conversations, export and recall them, to favourite prompts, and for knowledge sharing within your company and around the world.
How does OFS approach Mayvn AI responsibly?
02
Our goal is to empower users with tools that are safe, secure and as accurate as possible. We do that primarily by: → Selecting AI Models which are recognised to be reliable and consistent, to have a strong task-focus, with robust guard-rails against harmful content. → OFS does not conduct primary training of AI models, and we don’t use customer data for any training, testing or validation of AI models. → Providing strong privacy controls around data access – Mayvn AI systems cannot directly access any data, instead the user’s account is used to limit the reach to information that user already has access to via existing tools. → Presenting AI generated information with appropriate context – our user interfaces indicate that AI features make recommendations only, and we will endeavour to prompt users with questions they can follow-up with that allow for testing and validation of AI results with new data.
Things to consider about AI Generated Content
03
Generative AI is not capable of producing responses which are 100% accurate. While the models used by Mayvn AI are trained on vast datasets for accuracy, you should be aware of this inherent limit and validate all AI generated content for validity by interrogative prompting and using external data. It is an inherent limitation of Generative AI models that numerical outputs may not always be precise or consistent. While Mayvn AI is designed for high accuracy, users should be aware of potential limitations, including: → Approximation & Rounding – AI-generated figures may involve rounding errors or approximations. → Context-Dependent Interpretation – The accuracy of numerical responses depends on input clarity, assumptions, and data context. → Source Variability – When using external data sources (e.g., in data-based responses), discrepancies may arise due to dataset differences or versioning. Generative AI incorporates randomness for diverse responses, leading to variability: → Inconsistent Outputs – Identical inputs may yield different results. → RAG Variability – Retrieval-Augmented Generation (RAG) may produce differing responses to repeated queries. → Designed for Adaptability – Variability enhances robustness, avoiding rigid or repetitive outputs.
Security of our data when using Mayvn AI
04
Mayvn AI is designed to operate in a fully tenant-isolated way, built on AWS Bedrock’s certified platform for use of Generative AI Models. Before you interact with Mayvn AI: → Mayvn AI and AWS Bedrock have no access path to data held by OFS at customer level. → No information is sent to any AI systems or AWS Bedrock about you, your data, or your session. When you submit a prompt to Mayvn AI: → A stateless AI model is invoked and responds to your question. It may request a small extract of the available OFS data to complete a more detailed analysis, which is gathered in your web browser and sent by OFS software to the AWS Bedrock authenticated and encrypted endpoint, via a non-public connection. → No customer data is saved in AWS Bedrock, or outside your existing environment and home region to facilitate this interaction. → Some logging information, stripped of sensitive information, is securely captured to allow OFS to quantify usage of Mayvn AI, to measure themes, trends and identify usage patterns with new features. Each interaction with Mayvn AI is protected by the design choices OFS have made, including selection of the AWS Bedrock platform, choosing safe, enterprise-grade models, and protection of your data with encryption and authentication. At no point is any Mayvn AI data or customer data intentionally made accessible to 3rd parties, including AWS and AI Model Providers.
We self-host our OFS installation. Is anything different?
05
When your OFS installation is self-hosted, and runs on a server you control, or in your corporate IT environment, the security and safety features of the Mayvn AI design provide you with the knowledge that your data remains in safe hands. To start using Mayvn AI, your OFS software requires an upgrade. As such, no data leaves your environment until you have approved this upgrade. As part of the process of upgrading your software, your OFS representative will walk you through the new connectivity requirements that Mayvn brings to your OFS environment. Only an OFS-managed implementation of Mayvn AI is available, so it is not possible to run Mayvn AI in your own AWS account, should your organisation have one. Mayvn AI is only compatible with operating in AWS Bedrock, necessitating an outbound internet connection from your OFS Server (but not end user devices) to AWS Bedrock endpoints. → The connection between your self-managed OFS install and the AWS endpoint is authenticated and encrypted at all times. → Any data sent to Bedrock to use Mayvn AI is temporary and never saved by AWS Bedrock. → Some metadata and logging information is kept for the purpose of securing the OFS-managed AWS environment used to power Mayvn AI features.
I have further questions...
06
If you have questions that we haven't been able to answer here, we're more than happy to help. Simply reach out to your OFS representative, or contact us via the Contact form linked below, and our team will connect you with the right members of the Mayvn AI team to address any questions which remain. Our team has extensive documentation and resources which will help to communicate the architectural features of Mayvn AI which promote safety, accuracy and security.