Contents

Frequently asked questions

Mohammed Jamal Updated by Mohammed Jamal

This article answers common questions about using OneAdvanced AI. We aim to help all users maximise our UK sovereign generative AI to support their work, and understand you may have queries.

Please be aware we are always updating and improving OneAdvanced AI. Check back to this article to see the most up to date information. 

How do you use my data?

We do not use any prompts or customer data to train our AI model. Nor do we store this information.

The data you provide, and the context you put around it for your requirement, is provided solely for the interaction with OneAdvanced AI. You maintain full control of the data sources used by way of temporary file upload in to the chat, or permanent file uploads in Private Spaces (*Coming soon, see below).

The prompt that you create will reference to the data source (file) to determine the response (output). This ensures you remain firmly in the loop throughout the chat interaction.

What is the maximum file size that I can upload?

You can currently upload files that are up to 2MB in size.

Can I upload multiple files to use with a query?

Each file that you upload can be up to 2MB.

What file types am I able to upload?

We currently allow the following file types.

File Type Extension

File Type Name

Description

.csv

Comma Separated Values

A plain text file used to store tabular data, where each line represents a record and values are separated by commas. It's commonly used for importing and exporting data between applications and databases.

.docx

Word Document

A file format used by Microsoft Word to store word processing documents, including text, images, and formatting. It's a zipped archive containing XML files that define the document's structure and content.

.pdf

Portable Document Format

A file format used to represent documents in a self-contained, platform-independent format. PDFs preserve the layout and formatting of the original document, making it easy to share and view on different devices.

.pptx

PowerPoint Presentation

A file format used by Microsoft PowerPoint to store presentations, including slides, text, images, and multimedia content. It's a zipped archive containing XML files that define the presentation's structure and content.

.txt

Plain Text File

A simple text file containing unformatted text, often used for storing notes, logs, or configuration files. It's a plain text file with no formatting or binary data.

.xlsx

Excel Spreadsheet

A file format used by Microsoft Excel to store spreadsheets, including tables, formulas, and charts. It's a zipped archive containing XML files that define the spreadsheet's structure and content.

Can I upload .doc, .ppt, and .xls files?

No, these file types are not currently supported. Only the file types listed in What file types am I able to upload? can be uploaded.

Do you run a virus scan on the files we upload?

To ensure the security of the service, virus scanning is performed before a file is uploaded. When the chat with the temporarily uploaded files is deleted, so are the temporary files.

Do you have a Fair Usage Policy?

OneAdvanced AI should be used in accordance with our Fair Usage Policy. To read more, select the button below.

What are Private Spaces? (*Coming soon)

Private Spaces provide a secure environment for uploading your files.

Private Spaces offers a secure storage area for individuals (with user only access) and organisations (where only authorised users are given access). Data saved in a user's Private Space remains accessible only to that user, while organisation-wide data is restricted to authorised users within the same organisation on OneAdvanced AI.

This dedicated digital space ensures data safety through strong encryption, access controls, and multi-factor authentication, allowing you to leverage AI on your documents without security risks.

Future enhancements will enable team-based Private Spaces for controlled collaboration.

How is my data protected and stored?

Your data is fully encrypted end-to-end.

Our UK-based private hosted AI model ensures that your data remains in the UK, under your control and in compliance with the applicable data protection legislation (e.g. UK GDPR).

How do you protect end users?

Your organisation uses our single sign-on system, OneAdanced Identity to manage access into the AI. This provides a unified, convenient and secure access to any products in the OneAdvanced portfolio.

To read more about Identity, select the button below.

What accessibility standards does the AI adhere to?

OneAdvanced AI complies with WCAG 2.2 AA standard.

When should OneAdvanced AI not be used?

OneAdvanced AI should not be used for the prohibited uses outlined within the Acceptable Use Policy. To read more, select the button below.

How does OneAdvanced monitor the AI?

We use the OneAdvanced Platform Observability to monitor system health, proactively identify issues, optimise performance, ensuring that our system is running smoothly, efficiently, and effectively, providing the best possible experience for our users.

A known list of 1,000 prompts is used to determine model consistency with previous runs. When system is live, usage is monitored through application logs and system metrics - including the number of blocked requests due to breaching acceptable use guards.

What AI model do you use?

We are deeply committed to developing and using AI responsibly.

Guided by EU AI Act, ethical principles and compliance standards, we ensure our AI technologies are built with transparency, fairness, and a focus on societal and environmental benefits. With that in mind, fairness and bias mitigation were key factors in our assessment of shortlisted open-source AI models to provide to our customers for OneAdvanced AI. Meta has incorporated advanced techniques for bias mitigation in Meta Llama 3, ensuring that the model generates more balanced and neutral content which is crucial for the foundation of OneAdvanced AI. This includes fine-tuning the model to minimise biased outputs and creating systems to review and address any potential biases with use. Along with bias mitigation, OneAdvanced AI has implemented a range of guardrails and content filters to ensure that our responses are safe, respectful, and accurate.

On 12 December 2024, we signed the European Commission’s AI Pact, solidifying our commitment to ethical AI practices.

If you would like to read more about the LLM we use, see Meta's blog post here.

What is an LLM?

LLM stands for Large Language Model. It's a type of artificial intelligence (AI) designed to process and generate human-like language. LLMs are trained on vast amounts of text data, allowing them to learn patterns, relationships, and context, which enables them to understand and respond to natural language inputs.

What is Llama Guard 3?

Llama Guard 3 is a series of content safety classification models developed by Meta AI. It's designed to help ensure the safety of Large Language Model (LLM) applications by identifying and classifying potentially harmful content in both user inputs (prompts) and AI-generated outputs (responses). Through our Platform Observability layer, we can be flagged guardrail violations to ensure appropriate use of OneAdvanced AI.

Can I submit feedback?

Customer feedback is crucial to ensure we keep improving OneAdvanced for the benefit of all our users. Our users can submit feedback directly via the Product Hub menu within OneAdvanced AI, allowing our teams to assess and respond.

How did we do?

Contact