The “GenAI Gateway” platform, powered by AWS, is launched by the Digital Unit of BT Group, hastening the company’s safe and widespread deployment of generative AI.
Built in partnership with AWS, the GenAI Gateway is a generative AI enablement platform that gives BT Group safe and secure access to Large Language Models (LLMs) from a scalable number of foundation model suppliers, allowing them to harness the potential of GenAI.
The platform facilitates timely security, central privacy controls, use case-based charging, enterprise search, and the use of diverse corporate data sources, allowing the organization to exercise flexibility and accountability when implementing the many AI models that are necessary. Model exploitation risk is reduced by incorporated safeguards that promote ethics and performance (also known as “jailbreak risk”).
BT Group
As BT Group expands and speeds up the use of generative AI, a single, unified Group platform minimizes duplication of work and resources since APIs, security settings, infrastructure management, etc., can be controlled centrally, lowering the risk of mistake along the way.
The Digital Unit of BT Group has declared the opening of a cutting-edge internal platform that would enable the organization to use large language models (LLMs) from suppliers including Anthropic, Meta, Claude, Cohere, and Amazon. The GenAI Gateway is a vital tool that BT Group will use as it integrates AI into its operations. It was developed in partnership with AWS and makes use of AWS Professional Services, Amazon Bedrock, and Amazon SageMaker to provide secure, private access to a variety of natural-language processing and large language models.
While ad hoc use of LLMs is acceptable for testing and development, it is not a good fit for large-scale implementation; greater attention to cost control, security, and privacy is required. Additionally, LLM performance must be watched for unanticipated mistakes (such “hallucinations”) and model degradation over time (the point at which LLMs cease acting in a predictable manner). In the event that further problems arise, the GenAI Gateway also protects BT Group from being “locked in” to a certain LLM. Because the GenAI Gateway platform allows for per-use case budget monitoring, it will incentivize BT Group engineers to choose the appropriate model for the given use case at the most competitive price.
By consolidating platforms, BT Group can save redundant effort and resources while expanding the use of generative AI. Centralized administration of infrastructure, security configuration, and application programming interfaces (APIs) lowers the chance of mistake and the expense of keeping different LLMs for each use case.
The AWS-deployed GenAI Gateway, like every element of BT Group’s modular digital architecture, can only be accessed via secure APIs.
GenAI Gateway leverages two fully managed services
Amazon Bedrock, which provides a single API access to a selection of high-performing foundation models from top AI firms such as AI21 Labs, Anthropic, Cohere, Meta, Mistral AI, Stability AI, and Amazon; and Amazon SageMaker, which combines a wide range of tools to enable high-performance, low-cost machine learning for any use case.
Enterprise search, chat history, FinOps charging by use case, fast security, and the use of many corporate data sources are all supported by the platform. Central privacy measures secure data in accordance with Group policy and applicable regulations. These include distinct tenants for each use case, the use of Personal Identifiable Information filters, the placement of the data inside the UK, and the isolation of trained models from one another.
The GenAI Gateway has built-in guardrails that reduce the possibility of jailbreaks and harmful interactions. These guardrails block away queries that are not relevant to a particular application, ensuring that ethical and performance constraints are included from the beginning.
It will also use the “data fabric” data management platform to help enforce governing policies for how data can be used, as well as to manage access control and data sovereignty restrictions. GenAI Gateway is one of several key enablers AWS are deploying to enable BT Group as an AI-enabled enterprise.
The first beta use cases for GenAI Gateway are available right now. A pilot being conducted by Openreach aims to streamline procedures and increase productivity for its teams and clients of communications providers by compiling technical notes on Ethernet and full fiber installations. There is also a live second use case that supports contract analysis for the business, legal, and procurement departments of the group.
Businesses can employ generative AI rapidly and effectively at scale with the BT Group GenAI Gateway. Collaborating and working backwards from the client to create a mechanism to speed the deployment of generative AI use cases into production with integrated security and compliance has been a fantastic, innovative opportunity. The generative AI adoption flywheel effect will be sparked by the GenAI Gateway, giving BT Group and its clients faster outcomes.
AI is assisting it in rethinking the company’s future. So think that in cases when your data is consistent, the LLMs must be flexible. People can now access this formidable new suite of technologies at scale in a secure, responsible, adaptable, and scalable manner with GenAI Gateway, fulfilling the objective of using AI to unleash the potential of every person inside the BT Group both now and in the future.
