Want to learn more about GEMS AI and get hands-on practice? Explore our free GEMS AI eLearning on GEMS Academy!
Introduction
GEMS AI is a collection of products and features which include artificial intelligence capabilities. We are currently exploring and developing AI throughout GEMS but have some initial use cases which are now available to clients. These AI capabilities currently include:
AI Support Bot: Get bespoke support via a chatbot with instant answers to any of your support queries and links to further training documentation.
AI Document Analysis: Use the power of AI to query, compare, translate and summarize your documents.
AI Document Tagging: Save time by having AI automatically apply appropriate tags to documents uploaded to GEMS.
AI Autofill: Automatically extract data from a source document and have it entered in compatible GEMS Screens.
AI Analyze: Perform data analysis using the reasoning capabilities of GEMS AI and quickly understand risks, find missing datapoints, and compare data coverage across entities.
AI Entity Search: Quickly get insights or generate searches from your GEMS database using the power of natural language searching.
GEMS AI appears via contextual nudges within your existing GEMS platform leveraging AI technology. These capabilities are powered by third-party developed models that have not been exposed to any customer data to train the model, but have been configured to understand the structure and workflows of GEMS and its databases. We are continuously developing these capabilities and working on new use cases as these AI models improve over time, all the while ensuring that client data is secure and segregated from training the underlying model.
How does GEMS AI work?
GEMS uses various off the shelf third party large language models (LLMs) hosted by Microsoft Azure in a region aligned with your GEMS server region. Currently this is limited to LLMs from OpenAI such as ChatGPT but this may be expanded in the future as new models become available or to suit specific use cases as they are developed.
AI Support
GEMS AI Support provides personalized first-line support to common user questions. The AI support bot has been trained on GEMS user documentation with some help from our support team to provide full coverage of commonly asked questions.
The chatbot has been specifically designed to only answer questions from its knowledgebase of GEMS-specific documentation and it does not have any access to your underlying GEMS data.
AI Document Analysis, Document Tagging, and Autofill
GEMS AI Document Analysis and Tagging streamline how documents are handled within the GEMS platform by embedding intelligent automation directly into familiar user interfaces. Users are able to summarize, translate, and compare documents with a single click from key launch points such as Document Search and the Home Hub.
AI capabilities also support autofill—automatically extracting and inputting data from source documents into GEMS Screens.
While GEMS AI enhances productivity by automating document workflows and surfacing insights, it’s important to approach AI-generated content with care. The underlying models are designed to assist—not replace—human judgment. Outputs may occasionally contain inaccuracies, misinterpret context, or omit critical details. Always review AI-generated summaries, autofilled data, and suggested tags thoroughly before taking action or finalizing records. Translation functionality is not designed to replace professional translation and should not be relied upon to be accurate.
AI Analyze
GEMS AI Analyze leverages a "reasoning" LLM to analyze your data, letting you increase data accuracy, understand risks, and enhance your entity compliance capabilities.
AI Entity Search
GEMS AI Entity Search allows users to create data reports (searches) using their natural language. They can easily build searches for screens and data, summaries or ask for analysis on data within their GEMS database.
We use the AI LLM to convert the user's natural language request into something the system can understand (an API request) and then query the GEMS database and present the data back to the user. For the creation of large reports this information is then presented back to the user in tabular format and the data is not processed via the LLM. Where processing is required, for example if a user asks for a summary, average, or total, then the necessary information is provided to the LLM and a summary is returned. Please see below for more our data and privacy practices in relation to AI.
Data security
We understand how important data security is to our clients and our development process has been designed to continue to provide unrivalled privacy and security for your GEMS data.
GEMS AI Security and Policy Provisions
Proprietary Development
GEMS AI interface code is developed exclusively by Computershare and does not incorporate third-party or open-source code.Model Usage
No public AI models are utilized within GEMS AI. All AI functionality leverages secure Azure OpenAI services operating under an opt-out configuration, with access restricted solely to GEMS.Client Data Handling
GEMS AI does not train on client data. Client data is not retained beyond temporary processing cache and is not used for any model training purposes.Public Model Restrictions
No GEMS AI usage, including client data processing, utilize public models.Regional Data Residency
AI services are deployed in regional instances to ensure data remains within the GEMS-hosted region in EMEA or the United States.Access Controls
GEMS AI adheres to existing GEMS permission structures and will not permit access to data restricted by established access levels.Monitoring and Logging
Computershare maintains logs of AI activity and has roles responsible for monitoring and reviewing outputs for hallucinations, errors, and potential abuse.Compliance and Certifications
Azure OpenAI services comply, where applicable, with industry standards and certifications, ensuring GEMS AI meets stringent security and privacy requirements.Transparency of AI Usage
GEMS AI is fully transparent regarding its presence. Users interacting with GEMS AI through chat or other features will receive clear notification that they are engaging with an AI system.Governance and Impact Assessment
Computershare’s governance framework mandates AI Impact Assessments prior to enabling any new GEMS AI features. An inventory of all AI capabilities developed is maintained.
FAQs
Will my data be used to train AI models?
No, Computershare and its subprocessors do not use customer data to train any models and have specific agreements in place to ensure that prohibit this.
Does GEMS AI respect my users' Access Levels?
Yes, GEMS AI will respect your existing user Access Levels. Any data or responses that are fed back from the chatbot or GEMS AI will be filtered to respect any data visibility or access requirements that have been set for the specific user.
Does GEMS AI process documents automatically using AI?
No, documents workflows are only performed at the point the user selects to use the feature within GEMS. Documents will not be processed unless a user has expressly used the functionality and selected or provided a document or text.
Does GEMS AI retain copies of the documents / text I process using the document workflows?
No, AI document output is not retained. All documents which are processed by AI are retained for a 48 hour period for abuse monitoring purposes.
How to access GEMS AI?
The GEMS AI Assistant is located at the bottom right of your GEMS workspace.
