AI agents just rediscovered the dual-write problem from distributed systems
blaizedsouza · x · 2026-09-27
Raul Junco walks through a classic agentic failure mode: a support agent calls the payment provider, the refund succeeds, but the agent crashes before recording it. On recovery, the workflow retries and refunds the customer twice.
Key points:
- The core issue: agents can restore their reasoning state without knowing what happened in the real world outside the workflow.
- Conversation history is only one layer of agent state — there's more to persist than chat logs.
- Long-running execution frameworks (e.g. durable workflows) solve only half the problem; Junco outlines the safeguards to add, the decisions never to delegate to the model, and the invariant to test before shipping.
A valuable read for anyone building agents that touch payments or other irreversible side effects.
More from coding & agent
- Claude-generated Three.js 3D harbor scene wows devs with its detail — EricBuess · 2026-09-27
- Agent monitors flight cancellations for pennies using Mercator travel API in Grok Bot — jeff_weinstein · 2026-09-27
- Claude Opus 5.5 writes a 6:54 film as code, GitHub Actions renders the MP4 — EricBuess · 2026-09-27
- One prompt, 321 lines: Opus 5.5 builds a playable flight simulator in a single HTML file — EricBuess · 2026-09-27
- Claude Opus 5.5 one-shots a full retro Diablo-style RPG with graphics and audio, no generators — EricBuess · 2026-09-27
- Mnemos MCP author preps V3 rework: no setup, no multi-call turns, full engine complexity — RileyRalmuto · 2026-09-27