AWS Telecommunications United Kingdom Company newsroom

BT Group

Document Processing · Enterprise Knowledge Search

BT Group needed a secure, centrally governed way to scale generative AI beyond ad hoc experimentation, with cost control, privacy, prompt security, model monitoring, and flexibility to use multiple LLM providers without lock-in.

Reported impact

Business outcome

BT Group launched GenAI Gateway into live beta use, giving the company a centralized platform for safe, flexible generative AI adoption at scale. The first use cases include summarizing Openreach engineering notes to simplify processes and boost productivity, plus contract analysis for business, legal, and procurement teams.

  • Productivity Improvement
  • Risk Reduction
  • Scalability

Solution

BT Group’s Digital Unit built the internal GenAI Gateway with AWS as a shared generative AI enablement layer for business teams and engineers. The platform is deployed on AWS and exposed through secure APIs inside BT Group’s modular digital architecture so applications can access approved LLM capabilities through one controlled entry point instead of standing up separate model stacks. GenAI Gateway uses Amazon Bedrock to provide access to multiple foundation model providers through a single API and Amazon SageMaker to support machine learning operations for BT’s use cases. The platform adds prompt security, chat history, enterprise search, per-use-case FinOps billing, and access to multiple corporate data sources. BT also applies central privacy controls such as separate tenants per use case, PII filters, UK data location, and isolation between trained models. Guardrails filter unsafe or out-of-scope prompts, while BT’s data fabric platform enforces data governance, access control, and data sovereignty restrictions.

Data flow

Users and internal applications call GenAI Gateway through secure APIs. The gateway applies prompt security, privacy controls, tenancy separation, PII filtering, guardrails, and budget controls, then routes requests to foundation models accessed via Amazon Bedrock and AI tooling in Amazon SageMaker. Corporate data sources and enterprise search features are brought into the workflow, and BT’s data fabric governs policy enforcement, access, and sovereignty constraints before responses are returned to business use cases.

Solution architecture

4 components · 4 layers
  1. Compute
    • Amazon SageMaker It supports BT Group’s machine learning workloads as part of the gateway platform’s AI tooling.
  2. Orchestration
    • GenAI Gateway BT Group’s internal platform provides a central, secure API layer for accessing and governing generative AI use cases across the company.
  3. Governance
    • data fabric BT Group uses its data management platform to enforce data-use policies, access control, and data sovereignty restrictions.
  4. Uncategorized
    • Amazon Bedrock It supplies BT Group with API-based access to multiple foundation models from different providers through a single managed interface.

Architecture clues

  • Accessed via secure APIs as part of BT Group's modular digital architecture
  • Applies PII filters and keeps data within the UK
  • Built-in guardrails limit jailbreak and toxic interaction risk
  • Deployed on AWS
  • Includes per-use-case FinOps billing and budget tracking
  • Isolates trained models from each other
  • Live with first beta use cases
  • Supports multiple foundation model providers through a single API
  • Uses separate tenants for each use case

Evidence from the source

"A trial in Openreach is summarising engineering notes on Ethernet and full fibre jobs, helping to simplify processes and boost productivity"
"Ad-hoc use of LLMs, whilst appropriate for test and development work, is not well suited to large scale use; cost control, security and privacy need more careful management."
"Central privacy controls include separate tenants for each use case, the use of Personal Identifiable Information filters, the location of the data within the UK"
"GenAI Gateway, deployed on AWS, is accessed via secure APIs, like all the components of BT Group’s modular digital architecture."
"Guardrails are built into the GenAI Gateway, limiting the risk of jailbreaks or toxic interactions"
"The GenAI Gateway, built in collaboration with AWS and using Amazon Bedrock Amazon SageMaker and AWS Professional Services capabilities, provides secure, private access to a range of natural-language processing and large language models"
"The platform supports prompt security, chat history, FinOps billing per use case, enterprise search, as well as use of multiple corporate data sources."
"we will also use our ‘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."
Transparency Evidence & methodology 75/100 reference value · 92% confidence · 1 source

Review how this reference was scored and trace each normalized claim back to its documented source.

01

Scoring methodology

How scores are calculated

Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.

Customer reference value75/100
  • Problem clarity20/20
    Measures whether the customer's operational problem is specific enough to compare.
  • Implementation depth25/25
    Rewards named components, data flow, integrations, and architecture detail.
  • Outcome specificity20/20
    Measures whether the reported benefit is concrete rather than a generic claim.
  • Quantified impact0/25
    0 metrics describe measurable impact.
  • Deployment scale10/10
    Rewards evidence of a live or scaled deployment and meaningful integration scope.
Evidence confidence92%
  • Source authority25/25
    The deployment is described by the adopting organization or a government source.
  • Claim traceability21/25
    10 of 12 key fields link to a named source.
  • Direct evidence20/20
    8 direct source quotes retained.
  • Date provenance10/10
    The publication date comes from the source page or publisher.
  • Source transparency8/10
    1 source reference document publisher, URL, authority, retrieval, and supported fields.
  • Implementation specificity8/10
    Rewards source-backed components, data flow, quantified impact, and a specific outcome.
02

Traceability

Sources & provenance

Field coverage shows which facts each source supports. Dates are shown only when documented by the source.

  1. BT Group

    BT Group customer story

    First-party
    https://newsroom.bt.com/bt-groups-digital-unit-launches-genai-gateway-platform-powered-by-aws-accelerating-the-companys-safe-adoption-of-generative-ai-at-scale/
    Source type
    Customer-published story
    Published
    Sep 24, 2024
    Supports
    • Customer
    • Business problem
    • Solution
    • Products
    • Solution components
    • Data flow
    • Integrations
    • technical area
    • use case category
    • Business outcome
    • outcome category
    • Metrics
    • Architecture
    • evidence quotes
    • Publication date

Date evidence

  • Published on source page Sep 24, 2024 · BT Group · First-party
    Open date source
Not separately sourced Industry, Region, Company size. Treat these normalized attributes as lower-confidence until a dedicated source is added.