Memory vs History
Is keeping the entire conversation the same thing as remembering?
Compare full transcripts, rolling summaries and explicit task state across a 25-turn agent workflow.
Compare a full transcript, rolling summaries and explicit task state across a long-running agent workflow. Watch token usage grow while the information that actually matters changes much more slowly.
A scripted 25-turn task with three memory policies. No live model is called.
Related research
This archive contains independent research, experiments, architecture studies, and technical observations. Some findings are exploratory and should not be interpreted as peer-reviewed scientific conclusions unless explicitly stated. Architecture studies of third-party products are based on publicly observable behaviour, published technical information, and engineering analysis. They are not claims of access to private source code or infrastructure.