LLM's as a technology - currently - are stateless. The memory layer is controlled by the agent. I m surprised with the gpt-6 reference unless it has to do with vertical integration between the agent and the layer.
replyYeah, I am aware of this statelessness.
replyThis is what I was referring to:
https://www.cnbc.com/2025/08/19/sam-altman-on-gpt-6-people-w...
It's marketing speak, but the goal is clearly there, no idea how achievable.
How it works under the hood is separate from selling the feature.
reply