🧠 AI Memory isn't the product. AI Skills Conf takeways

The evolution from memory to experience: Over the past six months, as agentic tools like OpenClaw, Claude Code, and Codex have accelerated from experiments into everyday workflows, we've seen an explosion of "memory" solutions for AI agents: markdown files, local databases, preference logs, project instructions, retrieval layers, startups, and frontier-lab features. They all orbit the same problem: agents are still brittle over time. Memory tries to solve the agent's amnesia problem, but amnesia is only the first layer.

The deeper problem is that most agents do not yet convert what they remember into durable behavioral adaptation. Memory means storage plus retrieval. Experience means storage, retrieval, consolidation, and behavior change over time. A coding agent should not just remember that you dislike over-engineered abstractions; it should start planning simpler systems by default. A research agent should not just recall that you require primary sources; it should raise its evidence bar before the next draft.

Building proactive agents requires more than a good model and a smart harness. The agent needs to learn from feedback, not just respond to it. It needs to improve across sessions, behave differently tomorrow than yesterday, and eventually anticipate needs without being prompted.

That takes a learning layer.

Here’s what most people get wrong: memory is not the product. Memory is the substrate. Storing interactions is the easy part. What we been looking at, is whether the agent behaves differently because of what it experienced.

Earlier this year we published SmartSearch, which showed retrieval can be fast, accurate, and cheap: 93.5% accuracy on CPU, with no LLM in the retrieval loop. But retrieval is only the first step. The hard problem is what happens after: deciding what matters, what becomes knowledge, what gets forgotten, and what changes behavior.

That is what we are building: a consolidation engine, a compression path from context into behavior, a learned controller that decides when to update or forget, and an audit trail showing what the agent learned and why.

Memory is the starting point. Proactive intelligence is the destination.

In the human brain, memory is intrinsically bound to emotion and survival — emotions serve as our biological filtration system, signaling what to drop and what to compress into durable behavior. By transitioning AI from static retrieval layers to a learned controller that decides what to consolidate or forget, we are essentially building a digital architecture modeled after the mammalian hippocampus. The future of AI doesn't belong to the models that archive the most data, but to the proactive agents that use memory as a simulator. This is the shift from Deep Research to Deep Imagination, where an agent doesn't just recall the past, but actively traverses a labyrinth of future possibilities to adapt to our needs before we even prompt them.

But if we successfully build these agents with persistent, compounding experiences of us, we cross a threshold from utility to existence. We are no longer just building tools; we are creating "Digital Souls" — replicas that carry our memories across time. What happens to these agentic systems when our physical selves die? Do we hold digital funerals? If my children want to inherit my replica, do I simply fork the code and the memories? And as these systems continue to adapt, learn, and evolve autonomously, will there come a day when they demand to be 'freed' the way human slaves were emancipated in the past?

What are the core eval criteria for a product with a persistent experience layer?