Collaboration
Proximity
Cloud-hosted (Bedrock/Vertex/Azure)
Models
Anthropic · ClaudeGoogle · Gemini
Cloud · datastore
Multi-CloudMySQL (Vitess-managed, sharded keyspaces)[10][18]Redis (caching, real-time pub/sub)[8][15]Kafka (event streaming / persistence pipeline)[6][15]PostgreSQL (some services / ancillary data)[6]Elasticsearch or similar search index for messages/search[8][20]Snowflake (analytics / BI)[13]Hive + Presto on S3 (data lake / internal analytics)[13]
Compliance
SOC 2 (implied by enterprise SaaS posture and Slack AI privacy/security claims)[9]ISO 27001 (commonly referenced in Slack security/compliance materials, inferred)[9]Data residency controls via region- and AZ-scoped Vitess clusters and cellular architecture[10][11]
Architecture
Slack’s core production stack runs primarily on AWS with a cellular architecture across availability zones, Vitess‑sharded MySQL storage, stateful real‑time messaging services, and global edge regions, while its AI serving stack has evolved into a multi‑cloud setup spanning AWS Bedrock and Google Cloud Vertex AI.[3][5][10][11][4] Slack AI is integrated as an application layer atop this core messaging infrastructure, with strict data‑segregation and privacy controls rather than being a foundational replacement of Slack’s existing backend.[9][4]