Compressing Context
- Source
- factory.ai
- Date
Context
Factory keeps long coding conversations manageable by maintaining a saved summary of earlier work. When the conversation reaches a size threshold, it keeps recent messages verbatim and folds only the newly removed material into that summary. This avoids repeatedly summarizing the entire older conversation. Compression method.
What survives matters: Factory says the summary should preserve the session's intent and requirements, a record of files changed and test results, and breadcrumbs such as file paths that let the retrieve details again. Compressing too aggressively can erase information the agent soon needs, forcing it to re-fetch material and add extra calls that can outweigh the savings. Factory therefore aims to minimize per task rather than per request. Retention tradeoff.
- AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
- token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
- context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
Checking sign-in…
Loading comments…