Productivity · AWS
Notion
Notion’s core production architecture is centered on a sharded PostgreSQL system on AWS, with 96 physical Postgres instances and five logical shards per instance as of 2023. Its data lake pipeline uses Debezium CDC into Confluent Cloud Kafka, then Apache Hudi and S3 for downstream analytics, search, and AI-related workloads; the public sources do not confirm a native foundational model inside Notion’s core product.
Foundation proximity: Off the LLM map (in-house non-LLM ML)
Foundational models & how they consume them
Off the LLM map — in-house non-LLM ML only.
Cloud & datastore
AWSPostgreSQLAmazon RDSApache Kafka (Confluent Cloud)Apache HudiS3SnowflakeDebezium CDCSparkCompliance
data residency controls (US/EU regional processing)Similar companies
AirtableAsanamonday.comGrammarlyExplore
FAQ
Does Notion use a foundational AI model?
Notion relies on in-house, non-LLM machine learning at its core — off the foundational-LLM map.
What cloud does Notion run on?
Notion runs on AWS.
What database does Notion use?
Notion uses PostgreSQL, Amazon RDS, Apache Kafka (Confluent Cloud), Apache Hudi, S3, Snowflake, Debezium CDC, Spark.
How close does Notion run to the foundational model?
Notion is off the foundational-LLM map (in-house non-LLM ML).
Sources
notion.so ↗← Full orbit map (260 companies) · Score your own stack →
Architecture inferred from public sources · confidence medium · verify before betting on a detail.