ThirdBrAIn.tech

Advanced Video Search

AWS Data and AI Platform

3 min read

AWS Data and AI Platform

What it is

AWS has no single branded platform; the data layer is a set of services: S3 and S3 Tables (storage), Glue Data Catalog and Lake Formation (catalog and governance), Redshift/Athena/EMR (engines), SageMaker Unified Studio (workspace), Bedrock and AgentCore (AI), Amazon Quick (BI and agents).

Maker, ownership and history

Amazon Web Services. SageMaker launched 2017-11-29 (Wikipedia, secondary). The “next generation of SageMaker” bundles analytics and AI in Unified Studio (docs).

Editions/deployment

Managed services in AWS regions; consumption billing per service (AgentCore is consumption based, no minimums); Amazon Quick has Free and Plus plans and per-user subscriptions. Pricing pages are per service on aws.amazon.com.

Core architecture

  • S3 Tables: table buckets storing Apache Iceberg, including Iceberg V3 (deletion vectors, row lineage, variant), automatic compaction and snapshot management, IAM access via the s3tables namespace; integrates with the Glue Data Catalog for Athena, Redshift, Quick.
  • Glue Data Catalog and Lake Formation: central metadata, grants at database/table/column/row/cell level, LF-Tags and ABAC, cross-account sharing; federated catalogs unify S3 data lakes with Redshift and external sources such as BigQuery and MySQL; Iceberg REST compatible.
  • Redshift: warehouse (serverless option) queries Iceberg and can be registered in the catalog.
  • SageMaker Unified Studio: projects bringing SQL analytics, processing, model development and generative AI behind one interface.

Role in an enterprise AI rollout

Iceberg tables and catalog permissions feed RAG, agents and ML; Bedrock Data Automation turns documents, images, video and audio into structured output (with confidence scores and visual grounding) for RAG ingestion; AgentCore runs agents with governed access to data and tools.

AI features as of October 2026 (AWS docs)

  • Bedrock Data Automation: multimodal extraction API (document processing, media analysis, RAG enrichment).
  • AgentCore components: Harness, Runtime, Memory, Gateway (APIs/Lambda to MCP tools), Identity, Code Interpreter, Browser, Observability, Payments, Evaluations, Optimization, Policy (Cedar-compatible), Registry; works with LangGraph, CrewAI, Strands and MCP/A2A. Per-component GA/preview status not shown on the overview page.
  • Amazon Quick (evolved from QuickSight; Quick Sight remains a feature): chat, Flows, Automate, Index (grounding in company data), Research, Apps; MCP and OpenAPI connectors; Free and Plus plans.
  • Data-access governance for agents: Lake Formation row/column controls plus AgentCore Identity/Policy.

Integrations

Fabric mirrors AWS Glue Iceberg tables (microsoft-onelake-and-fabric-data-platform); Snowflake and Databricks read Iceberg on S3 (snowflake-ai-data-cloud, databricks-data-intelligence-platform). Google’s equivalent: BigQuery. Overview comparison: enterprise-data-ai-platforms-comparison.

Strengths and weaknesses vs peers (opinion)

  • Strong: primitives, breadth, native IAM, Iceberg-first storage.
  • Weaker: assembly required, many overlapping services (Glue/Lake Formation/Unified Studio), per-component maturity varies.

Self-learning

Sources (fetched 2026-10-07)

URLs above plus https://en.wikipedia.org/wiki/Amazon_SageMaker (secondary).

Open items

  • Redshift and Glue current-feature details not fetched (AWS What’s New page returned navigation only); no Redshift or AI-specific lakehouse announcements verified for 2026.
  • GA/preview status per AgentCore component and Quick plan contents unverified.
  • No current certification names verified.

Filter Videos

Tags

Advanced Tag Filters

Channels

Favorites

Active list: Quick Saves

This list is empty

Browse videos and save them to a list to get started

Recent Updates

See 1493 more →

Backlinks