Developer Demo Shows GPT-6 Astra Using a Persistent Memory Folder Across Tasks
Sign in✓ Signed in
A developer demo has put GPT-6 Astra's long-running agent memory in the spotlight, showing a custom workflow in which the model was instructed to read a persistent folder before each task and write back what it learned afterward. After 19 days, the experiment was described as having accumulated 165,000 leaves, 41,000 files and 118,000 calls, turning a simple file-based habit into a large external memory store.
Continue reading
The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in✓ Signed in — this article isn’t included in your current plan.
Unlocking the full article...