Discovery agents

Discovery Agents

What is the Offering? This bespoke service will help you build a set of Discovery AI Agents powered by AWS Bedrock, it is designed to de-risk, accelerate, and enrich Data Modernisation and Migration initiatives. At its core, the offering leverages Discovery Agents—AI-powered autonomous units capable of analysing metadata, system logs, usage patterns, reporting logic, and query activity across legacy data and reporting platforms. These agents:

Build end-to-end data lineage down to the attribute-level transformation logic. Extract and translate complex business logic into plain English. Identify and contextualise adhoc queries and usage footprints across systems. Uncover usage trends, business unit dependencies, and user-level impacts. Surface hidden risks, inefficiencies, and value opportunities early in the lifecycle. The solution is designed to run entirely on AWS utilising the following AWS services: Amazon S3 Amazon Bedrock Amazon DynamoDB Amazon OpenSearch Service Amazon ECR Amazon ECS / AWS Fargate Who is it for? The service is aimed at large, data-rich enterprises facing the challenge of modernising legacy environments. Key profiles include:

Primary Target Segments: CIOs / CDOs / Heads of Data Platforms looking to reduce risk and accelerate timelines in data platform transformation. Program Leaders managing enterprise reporting rationalisation, cloud migrations, or data warehouse modernisation. Data Governance / Architecture Teams seeking clarity over lineage and technical/business metadata. Business Leaders / Domain Owners wanting traceability of key KPIs and decision logic. How Does It Work? Discovery Agents: Core Mechanisms Uses a variety of metadata such as data platform activity logs, schema catalogs, ETL code, BI tools (e.g. SSRS, Tableau, PowerBI, MicroStrategy) report metadata. Automatically extract and unify metadata, usage telemetry, and lineage. Use LLMs and Agentic Reasoning to: Map lineage with attribute-level transformation logic. Convert code-heavy business logic into human-readable summaries. Cluster and interpret adhoc query behaviour across users. Identify report, dataset, and pipeline usage trends and redundancies. Interactive lineage visualisation dashboards. Usage profiling and impact heatmaps across teams and business units. Value Proposition Attribute-level Lineage - Eliminates ambiguity in source-to-target mapping Business Logic Translation - Speeds up stakeholder buy-in and cross-functional understanding Usage & Impact Analysis - Supports informed decisions on rationalisation and decommissioning Agentic Automation - Reduces manual effort and time-to-value Risk Identification - Uncovers hidden dependencies early, preventing costly surprises