Alkera

Alkera is an agentic platform for building, analyzing, and maintaining a data stack. It reads context from connected services and knowledge bases, keeping source, author, and review status for each entry, and traces column-level lineage across services so agents can check upstream dependencies and downstream impact before planning changes.

Agents detect warehouse and job failures, catch bad rows with SQL or Python tests, and repair failures using deterministic classification and LLM root-cause analysis guided by execution logs, lineage, knowledge bases, and code history. Fixes are tested in isolated environments against production data or custom inputs, with CI/CD gates publishing check results to GitHub and review and approval in Slack, GitHub, or Alkera's platform. Work can be planned in a graph and executed as notebook cells with per-cell compute, and repairs run on Alkera-hosted workers or locally in the desktop app. Alkera connects to services including Airflow, dbt, Fivetran, Snowflake, BigQuery, Databricks, Datadog, GitHub, Notion, Confluence, Slack, Looker, and Tableau.

It is aimed at data teams working across ingestion, transformation, warehouses, and analysis. Teams create an organization with preconfigured connections and permissions, with incidents and agent conversations shared at the project level and separate rights to view, steer, or approve changes. Agent actions such as applying a fix plan, testing, creating quality tests, merging, and backfilling each have Always ask, Auto, or Always approve settings, with Auto operating within admin-set limits.