What is AI Hub?
AI Hub is GlobalData’s central space for accessing our AI-powered research and analysis tools. It brings together Ava, GlobalData’s digital worker, and a Template Library of ready-to-use workflows designed to simplify and speed up your research process.
Key Components of AI Hub
Ava: Ava is GlobalData’s digital worker and AI Research Analyst.
Templates: They are ready-made research workflows built and verified by GlobalData’s analysts
Discover: Discover provides a collection of frequently asked questions to help you get started with Ava.
Why should I use AI Hub?
Accelerate Research: Let Ava automate repetitive and time-consuming research tasks.
Personalized Insights: Upload your own files for context and receive tailored answers.
Trusted Content: All insights are drawn from GlobalData’s verified proprietary data and analyst-curated content.
Deeper Analysis: Explore complex topics through structured templates and guided workflows.
Transparent Sources: Every answer links to its original source for full visibility.
Export and Share: Export Ava’s responses or share them easily with your team for collaboration and reporting.
Secure Environment: AI Hub operates with enterprise-grade data protection and confidentiality.
What is Ava?
Ava is GlobalData’s AI Research Analyst powered by agentic AI. It can interpret complex questions, search across proprietary structured and unstructured datasets, analyse uploaded documents, produce charts and tables, run multi-step workflows, and generate structured reports, always grounded in GlobalData’s verified content ecosystem.
How is the new Ava Digital Worker different from the previous Ava?
The upgraded Ava powered by agentic AI uses an orchestrated chain of reasoning steps to break down queries, route them through the correct search pipelines, select datasets automatically, and assemble structured analyst-style outputs. Compared to the previous version, now Ava plans multi-step workflows, provides transparent “Thoughts” and “Action Plan” steps, shows full source citations, and delivers more reliable, complete responses.
How does AI Hub integrate with my current GlobalData licences?
Ava only searches content from the Intelligence Centers and datasets that your organisation is licensed for. This ensures responses are fully scoped to your permitted access and grounded strictly in GlobalData’s verified proprietary datasets.
What is the difference between Ava, Templates, and Discover?
Ava: A conversational AI analyst for freeform questions, file analysis, and custom instructions.
Templates: Analyst-built, structured research workflows that generate full reports.
Discover: A library of expert-curated sample questions that users can personalise and run instantly.
What types of questions can Ava answer?
Ava can answer market questions, competitor intelligence, company-level analysis, strategic and thematic analysis, financial comparisons, regulatory insights, technology trends, and document summarisation. It can also produce tables, charts, structured reports, presentations, and email-style summaries.
Can Ava combine numerical and non-numerical analysis?
Yes. Ava routes each query through the appropriate search pipeline, semantic search for text-based content, or SQL/structured search for numerical datasets, allowing it to combine both into a single unified answer.
Can Ava run cross-dataset or cross-Intelligence Center analysis?
Yes, Ava can run cross-dataset queries within the same Intelligence Center. No, Ava cannot currently pull insights across multiple Intelligence Centers to merge findings.
Can Ava show the sources behind every answer?
Yes. Every data point and paragraph is linked to its original GlobalData source through numbered citations that open in Canvas View. This ensures traceability and auditability.
Does Ava hallucinate?
Hallucinations are minimised because Ava never searches the open web and relies exclusively on GlobalData’s verified proprietary content. Every answer includes citations so you can validate the information.
What file types can I upload?
You can upload PDFs, Word documents, PowerPoints, images (JPG, JPEG, GIF, TIFF, BMP), and text-based formats. Ava can extract, summarise, and analyse content from all supported files.
Is OCR supported for scanned or non-readable PDFs?
Yes. Ava includes OCR capabilities that allows it to read and interpret scanned or image-based PDF content.
How many files can I upload and what are the limits?
You can upload up to 10 files at a time. Files must be under 10MB each, with a total storage allowance of 2GB across uploads.
Can Ava analyse multiple files at once?
Yes. Ava can cross-reference and summarise multiple files within a single conversation, keeping full context throughout the session.
How can I export Ava’s responses?
You can export responses as Word (.docx), PowerPoint (.pptx), or PDF documents. Visuals can be exported as images or raw data formats like CSV/XLSX.
Does Ava remember previous messages?
Ava remembers your instructions and files within an active conversation, allowing you to refine and expand analysis naturally. Memory resets automatically when you start a new chat.
Can I delete my conversation history?
Yes. You can delete any chat and uploaded documents at any time from chat history. Deleted items are permanently removed from GlobalData’s systems.
Is there a limit to how many conversations I can have?
No. You can run unlimited workflows and conversations. During rare periods of high usage, some requests may temporarily queue.
Is there a limit to the historical chats I can access?
No. You can access your entire historical transcript list unless you manually delete conversations.
What is GlobalData’s AI infrastructure?
GlobalData has built a robust AI-ready infrastructure designed to make all its data seamlessly compatible with large language models (LLMs) and modern AI applications. Our AI infra-architecture is designed to unlock the power of our proprietary data assets.
GlobalData’s AI infrastructure is a combination of optimized data engineering, semantic search capability, AI powered query translation, and a scalable server setup. This enables efficient interaction between LLMs and business datasets, providing faster, smarter, and more accessible insights.
Data Compatibility
All textual data is processed by chunking, embedding, and indexing within an Elasticsearch database. This enables semantic search capabilities.
Datasets stored in SQL Server are transformed into simpler table structures, allowing LLMs to easily interpret and generate optimized queries for retrieving insights.
AI Query Layer
A custom layer has been developed on top of Elasticsearch to translate LLM generated prompts into DSL (Domain Specific Language) queries, making intelligent querying more efficient.
Infrastructure Deployment
The system runs across more than 70 servers, covering Vector database servers, ETL servers, Web servers and SQL database servers, ensuring high availability, scalability, and smooth data processing across applications.
Which tools make up GlobalData’s GenAI tech stack?
GlobalData’s GenAI technology stack is a combination of programming frameworks, AI platforms, cloud services, and databases:
Core Development
Backend: Python (leveraging multiple AI/ML SDKs and frameworks).
Frontend: .NET for building user facing applications.
Databases
Elasticsearch (Primary) and OpenSearch for semantic search and unstructured data querying.
MS SQL Server for structured data storage, transformation, and retrieval.
GenAI SDKs & Frameworks
Haystack for building search and Q&A systems.
LangChain for prompt orchestration and workflow automation.
OpenAI SDKs & Agents SDK for LLM integration and intelligent agent capabilities.
PromptLayer for managing and optimizing prompts.
Platforms & Cloud Services
OpenAI Platform as primary model provider and for hosting fine-tuned LLM.
Hugging Face Platform for open-source model access.
AWS and Azure for scalable cloud infrastructure, deployment, and compute resources.
The GenAI stack combines Python/.NET development, Elasticsearch OpenSearch MS SQL database layer, GenAI frameworks (Haystack, LangChain, OpenAI), and enterprise grade platforms (AWS, Azure) to power scalable and production ready AI applications.
What is GlobalData’s AI architecture and methodology?
GlobalData’s AI architecture is designed to deliver fast, accurate, and context rich answers by combining intelligent query routing with LLM driven processing.
Query Parsing & Routing: Every user query is first parsed and analyzed, then routed via the Ava engine to the most appropriate pipeline — Structured Search, Structured SQL, or Semantic.
Proprietary Databases: Responses are generated using GlobalData’s proprietary structured and unstructured databases as the core source of truth.
LLM Powered Processing: Within each pipeline, multiple LLM calls are orchestrated to interpret the query, extract the most relevant information, and summarize it into a clear, user-friendly answer.
Is GlobalData’s AI interoperable?
GlobalData is continuously enhancing interoperability in line with evolving market standards.
Current State: Our AI currently provides APIs that clients can connect with external systems and services for data access and integration.
Future Direction: We are transitioning from a RAG framework to the Model Context Protocol (MCP) — an open standard that enables secure, standardized connections between AI models and external da
What guardrails are in place against hallucinations?
Hallucination Mitigation Strategies:
Accuracy Evaluations:
Regular assessments use a pre-built library of questions to systematically review AI outputs, identify hallucinations, and inform improvements.
Prompt Engineering:
Carefully crafted, specific prompts guide the AI to produce accurate and relevant responses, reducing ambiguity.
Retrieval-Augmented Generation (RAG):
AI responses are grounded in factual data by retrieving relevant information from proprietary databases and documents.
Guardrails Against Non-Compliant Queries:
We are currently using OpenAI’s moderation API in all user-inputted query processing
What happens to the documents I upload?
Uploaded documents are stored securely in your personal account, accessible only to you. They are never used for AI model training and are automatically deleted after 45 days unless you delete them earlier.
Who can see my uploaded data?
Only you. GlobalData employees and other clients cannot access your uploaded content. Data is not used for training and does not leave the secure environment.
How do you ensure uploaded files are virus-free?
All files are scanned using automated antivirus and malware detection tools before they enter the system. Suspicious files are rejected or quarantined immediately.
What data is collected and processed?
The AI Hub AVA Search is a chatbot that allows users to submit queries in natural language or via pre-configured prompts. The user's query, along with any necessary metadata, is collected and processed to retrieve the requested data and information. The data is used solely to provide accurate responses and improve the user experience. Please note that we do not anonymise the data submitted through the chatbot.
Do you retain data to train your AI models?
No; customer data is not used to train our AI models. However, certain data may be retained for usage tracking and service improvement purposes. Data submitted during chatbot interactions is stored in accordance with our data retention policies and used solely for operational monitoring, performance analysis, and enhancing user experience. It is not shared with third-party providers for training or model development. Users have the option to delete their chat history at any time, supporting transparency and control over personal data.
Is any of the data shared with any third parties?
GlobalData uses the OpenAI API to support its large language model (LLM) functionality. Depending on how a user phrases their query, the input may be securely transmitted to OpenAI to generate a relevant SQL query or response. While no personal or sensitive data is intentionally shared, users are clearly instructed not to include any such information in their queries.
Use of AVA is subject to our Terms of Use and Privacy Policy, which require users to follow data protection best practices and avoid submitting prohibited data. This helps ensure continued GDPR compliance and responsible use of third-party services.
Do you share data with other third-party providers?
Apart from OpenAI, we may use other trusted third-party providers to help deliver the service, such as cloud hosting or analytics platforms. All third parties are contractually bound to comply with applicable data protection laws, including GDPR, and to ensure the security of any data processed.
Where is your data stored?
All data is stored in secure SQL databases hosted within Amazon Web Services (AWS) data centres located in Ireland.
What encryptions methods are used to protect data at-rest and in-transit?
Data is stored in SQL databases and encrypted at rest using Transparent Data Encryption (TDE) with AES-128. Data in transit is protected using Transport Layer Security (TLS) version 1.2.
How is data access secured?
The company follows the principle of least privilege, ensuring that users have only the minimum level of access required for their role. Access to data is strictly controlled using role-based access controls (RBAC) and enforced through multi-factor authentication (MFA). Only authorised
members of the development team and system administrators have access to the data, and all access is logged and regularly reviewed.
All access is logged, monitored, and subject to regular audits. Access rights are reviewed to ensure continued compliance with security and data protection policies.
Are you GDPR Compliant?
Yes. Data is collected and processed in line with GDPR. Yes. We are fully compliant with the General Data Protection Regulation (GDPR). All personal data is collected, processed, and stored in accordance with GDPR requirements. Data is securely hosted in Amazon Web Services (AWS) data centres located in Ireland, ensuring that it remains within the European Economic Area (EEA). We implement appropriate technical and organisational measures to protect personal data and support data subject rights under GDPR.
Does GlobalData follow the European AI act guidelines?
GlobalData is in full compliance with the EU AI Act. GlobalData does not develop and/or use AI tools and systems for prohibited practices and it’s AI tools are not deemed as high-risk systems. GlobalData’s proprietary AI tools and systems do not fall within the category of General Purpose AI Models. Notwithstanding the foregoing, GlobalData ensures its staff are adequately trained on the use of AI, taking into account ethical considerations when developing AI tools and systems.
Can I upload documents belonging to another company?
We strongly advise you to upload only documents you own or have permission to use. Users are responsible for ensuring compliance with IP laws.
What if my files contain confidential or sensitive information?
Uploaded files remain private to you and are not shared internally or externally. However, you are responsible for ensuring you have the right to upload and process the content.
Does GlobalData gain ownership of my uploaded content?
No. You retain full ownership of all uploaded materials. GlobalData does not obtain any rights or usage permissions beyond what is required to fulfil your query.
How accurate are the AI’s answers?
Ava’s accuracy is high because all results come from proprietary, verified GlobalData content. However, users should always validate key information. Source citations and Canvas View provide complete transparency.
What if Ava gives an incorrect answer?
You can refine your question, request clarification, or check the citations shown in the response. You may also flag the result using the thumbs-down feedback button for internal review.
What should I do if I experience issues?
If you experience issues with AI responses, use “Ask an Analyst” for content-related support. For technical problems, contact Client Support via the Help & Support section of your profile.
Is there a usage limit?
There is no fixed limit on workflows or conversations, though fair usage policies apply. During peak traffic, delays may occur.
Does cancelling my subscription affect my data?
Your data remains accessible until your subscription ends, after which it is deleted. You may export or delete your information before expiry.
Is there a mobile version?
AI Hub works on mobile browsers. A standalone mobile app is planned for future development.
Does it integrate with cloud storage platforms?
Not yet. Integration with services like OneDrive or Google Drive is part of the roadmap.
Can clients integrate with their own AI platforms to access our market intelligence data?
Yes, we can provide database access to client AI platforms via APIs or MCP servers. We can also offer Agentic AIs as API endpoints, covered under a separate DDS offering.