Absurd In Production

2026-04-04 · Source: Armin Ronacher's Thoughts and Writings · Field: Technology & Digital — Software Development & Engineering, Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure · Depth: Advanced, medium

Summary

Earendil has been running Absurd, a durable execution system built entirely on Postgres, in production for five months since its initial release on November 3, 2025. Absurd uses a single SQL file for task management, checkpoint storage, event handling, and claim-based scheduling, complemented by thin SDKs for TypeScript, Python, and Go. The system allows tasks to be decomposed into checkpointed steps, enabling retries from the last completed step and supporting long-running, suspended workflows. Key enhancements include decomposed `beginStep()` / `completeStep()` for conditional logic, task result fetching for child workflows, a CLI tool `absurdctl` for debugging, and a web dashboard `Habitat` for monitoring. The core design, emphasizing SQL-based complexity and thin SDKs (TypeScript SDK ~1,400 lines, Python SDK ~1,900 lines), has proven robust, particularly its checkpoint-based replay model and pull-based scheduling.

Key takeaway

For AI Architects or Machine Learning Engineers evaluating durable execution solutions, Absurd offers a compelling, minimalist alternative to complex external services. Its Postgres-centric design and thin SDKs simplify debugging and deployment, making it ideal for agent workflows and background processing that require fault tolerance. Consider adopting Absurd to reduce infrastructure overhead and streamline the development of resilient, long-running tasks, especially if you prioritize self-hosting and direct database interaction.

Key insights

Absurd demonstrates that durable execution systems can be built effectively and minimally on Postgres alone.

Principles

• Centralize complexity in SQL, keep SDKs thin.
• Checkpoint-based replay simplifies code determinism.
• Pull-based scheduling enhances self-hostability.

Method

Absurd manages durable workflows by registering tasks, decomposing them into checkpointed steps, and storing all state in Postgres, allowing retries from the last successful step.

In practice

• Use `beginStep()` / `completeStep()` for conditional workflow logic.
• Employ `absurdctl` for command-line debugging of stuck tasks.
• Integrate `Habitat` for real-time workflow monitoring.

Topics

• Durable Execution
• Postgres
• Workflow Management
• Absurd System
• Agent Workflows

Code references

• earendil-works/absurd

Best for: AI Architect, Machine Learning Engineer, Software Engineer, AI Engineer, MLOps Engineer

Related on AIssential

• Write Reliable Software — Durable execution ensures software reliability by automatically managing state and recovery from failures, abstracting distributed systems complexity.
• Microsoft Open-Sources PostgreSQL Extension for In-Database Durable Execution — "pg_durable" enables resilient, in-database SQL workflows, eliminating external orchestration for fault tolerance.
• How Stripe built “minions”—AI coding agents that ship 1,300 PRs weekly from Slack reactions | Steve Kaliski (Stripe engineer) — Stripe's "minions" are AI coding agents that significantly boost development velocity through automation and machine payments.
• Why Enterprises are Moving Critical AI Workloads On-Premise — Rising cloud costs, latency, and data sovereignty are driving enterprises to deploy critical AI workloads on-premises.
• Autonomous Database Agents: From Query Optimization to Self-Healing Systems — Autonomous AI agents are evolving to eliminate human decision latency in database administration, enabling self-healing systems.
• AI agents are entering their rebuild era as enterprises confront the reliability problem — Enterprise AI agent reliability in production hinges on robust workflow orchestration, state management, and recovery, not just LLM performance.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Armin Ronacher's Thoughts and Writings.