WhyLabs

aid: whylabs
url: https://raw.githubusercontent.com/api-evangelist/whylabs/refs/heads/main/apis.yml
name: WhyLabs
type: Index
image: https://kinlane-productions.s3.amazonaws.com/apis-json/apis-json-logo.jpg
tags:
- AI Observability
- ML Monitoring
- LLM Monitoring
- Open Source
- whylogs
- LangKit
- Discontinued
description: WhyLabs was an AI observability platform focused on data and model monitoring for both classical ML and LLM
  workloads. It built and maintained whylogs, an open-source data logging library that produces statistical profiles of
  tabular and unstructured data, and LangKit, an open-source toolkit for LLM telemetry covering relevance, toxicity, prompt
  injection signals, and quality metrics. WhyLabs, Inc. has announced it is discontinuing operations and has open-sourced
  its platform; the whylogs and LangKit projects remain available on GitHub for community use and research.
created: '2026-05-23'
modified: '2026-05-23'
specificationVersion: '0.19'
apis:
- aid: whylabs:whylogs
  name: whylogs
  tags:
  - Open Source
  - Data Logging
  - Profiling
  - Monitoring
  humanURL: https://whylogs.readthedocs.io/
  properties:
  - url: https://whylogs.readthedocs.io/
    type: Documentation
  - url: https://github.com/whylabs/whylogs
    type: SourceCode
  - url: https://pypi.org/project/whylogs/
    type: SDK
  description: whylogs is an open-source data logging library that creates approximate statistical profiles of datasets,
    enabling drift detection, data quality monitoring, and bias analysis for ML pipelines. Supports tabular, text, image,
    and embedding data and produces privacy-preserving profiles that can be shared and compared without exposing raw data.
- aid: whylabs:langkit
  name: LangKit
  tags:
  - Open Source
  - LLM Monitoring
  - Telemetry
  - Safety
  humanURL: https://github.com/whylabs/langkit
  properties:
  - url: https://github.com/whylabs/langkit
    type: SourceCode
  - url: https://pypi.org/project/langkit/
    type: SDK
  description: LangKit is an open-source toolkit that extracts telemetry from LLM prompts and responses including relevance,
    sentiment, toxicity, prompt injection signals, jailbreak similarity, refusal patterns, and quality metrics. Designed
    to plug into whylogs profiles for end-to-end LLM observability.
- aid: whylabs:whylabs-observability
  name: WhyLabs Observability Platform
  tags:
  - SaaS
  - Observability
  - Discontinued
  humanURL: https://whylabs.ai/
  properties:
  - url: https://whylabs.ai/
    type: Documentation
  description: WhyLabs Observability is the historical commercial SaaS that ingested whylogs profiles and LangKit telemetry
    for dashboards, drift alerts, and constraint monitoring. WhyLabs, Inc. has announced it is discontinuing operations
    and open-sourced the platform; commercial availability of the hosted service should be re-verified directly with the
    company.
common:
- type: Website
  url: https://whylabs.ai/
- type: GitHubOrganization
  url: https://github.com/whylabs
- type: GitHubRepository
  url: https://github.com/whylabs/whylogs
- type: GitHubRepository
  url: https://github.com/whylabs/langkit
- type: WhylogsDocumentation
  url: https://whylogs.readthedocs.io/
- type: LinkedIn
  url: https://www.linkedin.com/company/whylabs/
- type: CompanyStatus
  url: https://whylabs.ai/
- type: Features
  data:
  - name: whylogs Profiling
    description: Privacy-preserving statistical profiles of tabular, text, image, and embedding data.
  - name: LangKit LLM Telemetry
    description: Out-of-the-box metrics for relevance, toxicity, prompt injection signals, and refusal patterns.
  - name: Drift Detection
    description: Compare profiles over time to detect data and concept drift.
  - name: Data Quality Monitoring
    description: Constraint-based checks on schema, ranges, missingness, and distribution properties.
  - name: Bias and Fairness Analysis
    description: Profile-driven analysis of model inputs and outputs across protected groups.
  - name: Open Source
    description: Core libraries remain available under permissive licenses on GitHub.
- type: UseCases
  data:
  - name: ML Data Quality
    description: Monitor training and inference datasets for schema drift and quality issues.
  - name: LLM Telemetry
    description: Instrument LLM applications with LangKit metrics to track safety and quality over time.
  - name: Model Drift Monitoring
    description: Detect distribution shifts in features and predictions for production ML models.
  - name: Privacy-Preserving Logging
    description: Share statistical profiles between teams and environments without exposing raw data.
- type: Integrations
  data:
  - name: pandas
    description: Profile pandas DataFrames directly with whylogs.
  - name: Spark
    description: Generate whylogs profiles from PySpark and Spark Scala jobs.
  - name: Snowflake
    description: Profile Snowflake tables for drift and quality monitoring.
  - name: AWS S3
    description: Read and write whylogs profiles to S3 for distributed pipelines.
  - name: MLflow
    description: Log whylogs profiles alongside MLflow runs and models.
  - name: Hugging Face
    description: Apply LangKit metrics to Hugging Face model outputs.
maintainers:
- FN: Kin Lane
  email: [email protected]