Chalk
Real-time data platform delivering fresh features for AI inference decisions.
About
Chalk is a San Francisco data platform built for what it calls inference data pipelines: supplying fresh data with complex computation at the exact moment an AI model makes a decision. Where data warehouses and feature stores serve training and pre-computed data, Chalk targets low-latency inference use cases like fraud prevention, instant lending decisions, and content moderation.
The company raised a $50 million Series A in May 2025 at a $500 million valuation led by Felicis, with General Catalyst, Unusual Ventures, Xfund, and Triatomic Capital participating. Developers write features in pure Python, which Chalk compiles into efficient C++ and Rust pipelines delivering single-digit-millisecond latency at scale. Customers include Doppel, Sunrun, Whatnot, and MoneyLion.
Caveats: it is a young company in a space where Databricks and Snowflake are entrenched, and investor framing like the Databricks of the AI era is marketing rather than a measured claim.
What it can do
- Real-time inference data pipelines
- Python-defined features compiled to C++/Rust
- Single-digit-millisecond serving latency
- Online/offline feature consistency
- Streaming and batch sources
Setup
Sign up for Chalk Cloud or deploy in your VPC; define features in Python via the SDK.
Prerequisites
Data sources to connect; Python environment for feature definitions.
Use cases
Works with
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 82; Trending score 84/100 from researcher assessment.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.