What does this research mean for the field?

Hybrid memory outperforms experiential, baseline, and synthetic memory conditions in long-running AI coding agents. Novelty: ClaimNovelty.NOVEL_FINDING. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

The research aims to compare the effectiveness of different memory types in AI coding agents.

March 1, 2026Open Access

Experiential vs Synthetic Memory in Long-Running AI Agents: A Controlled Experiment in Software Development

Key Points

The research aims to compare the effectiveness of different memory types in AI coding agents.
Conducted a controlled experiment with four memory conditions.
Evaluated long-running AI coding agents under each memory type.
Measured performance based on a predefined scoring system.
Hybrid memory achieved the highest score (4.95/5.0).
Experiential memory scored 4.55.
Baseline memory scored 3.30 and synthetic memory scored 2.65.

Abstract

We present the first controlled experiment comparing four memory conditions in a long-running AI coding agent. Hybrid memory (4.95/5.0) outperforms experiential (4.55), baseline (3.30), and synthetic (2.65). v2.1: date correction (February 2026).

Read Full Paperexternally

Mark Helpful

Bookmark

Relay

View Full Paper