Architecture comparison

Block vs Klarna

Block runs closer to the foundational model (Direct API) than Klarna (Direct API). Shared foundation: OpenAI · GPT (highlighted).

Block

FinTech

Proximity

Direct API

Models

Anthropic · ClaudeOpenAI · GPT

Cloud · datastore

AWSMySQL

Compliance

SOC 2PCI DSS

Architecture

Cash App Money Bot + Square Managerbot run on Anthropic Claude (Sonnet) and OpenAI GPT via Block’s open-source goose agent framework (Claude is goose’s default on Databricks).

Klarna

FinTech

Proximity

Direct API

Models

OpenAI · GPT

Cloud · datastore

AWSAmazon DynamoDBPostgreSQLOracle DatabaseApache Kafka (as log/event store)Hadoop HDFSAmazon S3Amazon RedshiftRiakRabbitMQ (message broker)

Compliance

AWS financial services security and compliance alignment (core banking on AWS, collaboration with AWS compliance/security teams)[10]Bank-grade security controls inferred from licensed bank status and AWS FSI reference architectures[10]

Architecture

Production is built as microservices on AWS (including EKS/ECS) with event-driven flows via Kafka and a data platform spanning DynamoDB, S3, Glue/Spark, Hadoop/HDFS and Redshift, plus legacy Riak/RabbitMQ in some paths.[1][2][5][6][8][9][10][11] Klarna’s AI customer service stack layers agent orchestration (e.g., LangGraph) and RAG on top of OpenAI GPT-4-class models, integrated with internal APIs and Zendesk.[3][4][7]

← Full orbit map · Score your own stack →