Architecture

Agent-first engineering begins with human intent. Agents help translate that intent into systems through a deliberate sequence: Human intent → specification → architecture → implementation → verification → evidence → human review → deployment. Each stage produces something inspectable: requirements, design decisions, code, test results, deployment plans, and operational observations. Humans retain authority over consequential changes. Evidence makes that review concrete. The sequence also needs feedback. A failed test may expose a design assumption. A deployment plan may reveal a missing permission or an unclear ownership boundary. An operational result may show that the specification defined success too narrowly. ...

Karma now ingests, normalizes, and routes events from any CDC-like source into a shared ledger, optional graph, and an action loop — setting the stage for learned expectations and autonomous intervention.

In Karma, every action is just another event. This post explains the pattern for turning anomalies and rules into commands, tracking their execution, and feeding the results back into the same event pipeline.

Karma uses a single normalized event stream to feed both a ClickHouse ledger and an optional graph database — but through separate pipelines for flexibility, scalability, and clarity.