<?xml version="1.0" encoding="UTF-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
  <title>ebbmem Blog — Memory that knows what to forget</title>
  <subtitle>Engineering insights, architecture deep-dives, and technical analysis on persistent agent memory and verifiable data erasure from the ebbmem team at Spectrayan.</subtitle>
  <link href="https://ebbmem.ai/atom.xml" rel="self" type="application/atom+xml" />
  <link href="https://ebbmem.ai/blog" rel="alternate" type="text/html" />
  <id>https://ebbmem.ai/blog</id>
  <updated>2026-11-10T00:00:00.000Z</updated>
  <entry>
    <title>Two Kinds of “Dreaming” in Agent Memory: Offline Consolidation vs. Our Experimental Dream Pathway</title>
    <link href="https://ebbmem.ai/blog/experimental-dream-pathway-in-agent-memory" rel="alternate" />
    <id>https://ebbmem.ai/blog/experimental-dream-pathway-in-agent-memory</id>
    <published>2026-11-10T00:00:00.000Z</published>
    <updated>2026-11-10T00:00:00.000Z</updated>
    <author>
      <name>The ebbmem team at Spectrayan</name>
    </author>
    <category term="Architecture &amp; Infrastructure" />
    <summary>Why the AI agent ecosystem currently uses “dreaming” for two distinct mechanisms—offline transcript consolidation and generative hypothesis construction—and why our experimental Dream pathway enforces strict source honesty.</summary>
  </entry>
  <entry>
    <title>What We Plan to Explore Next: Hopfield-Style Associative Recall for Agent Memory</title>
    <link href="https://ebbmem.ai/blog/planned-associative-hopfield-memory-in-ebbmem" rel="alternate" />
    <id>https://ebbmem.ai/blog/planned-associative-hopfield-memory-in-ebbmem</id>
    <published>2026-11-10T00:00:00.000Z</published>
    <updated>2026-11-10T00:00:00.000Z</updated>
    <author>
      <name>The ebbmem team at Spectrayan</name>
    </author>
    <category term="Research &amp; Foundations" />
    <summary>A look ahead at why we plan to explore attractor-style Hopfield associative memory in ebbmem to complement store-and-retrieve indexing on partial cues—and the strict isolation, decay, and erasure invariants any future design must satisfy.</summary>
  </entry>
  <entry>
    <title>Inside the Engram Header: Cognitive Parameters and Source Honesty for AI Agents</title>
    <link href="https://ebbmem.ai/blog/cognitive-parameters-and-source-honesty-in-agent-memory" rel="alternate" />
    <id>https://ebbmem.ai/blog/cognitive-parameters-and-source-honesty-in-agent-memory</id>
    <published>2026-11-09T00:00:00.000Z</published>
    <updated>2026-11-09T00:00:00.000Z</updated>
    <author>
      <name>The ebbmem team at Spectrayan</name>
    </author>
    <category term="Architecture &amp; Infrastructure" />
    <summary>A technical tour of the cognitive signals specified on every trace in Memory Fundamentals (importance, retrieval strength D, storage strength S, valence, arousal, and open-loop state) and why closed provenance tags prevent agent confabulation.</summary>
  </entry>
  <entry>
    <title>Memory Fundamentals: A Formal Model for Memory Engines</title>
    <link href="https://ebbmem.ai/blog/memory-fundamentals-a-model-for-agent-memory-engines" rel="alternate" />
    <id>https://ebbmem.ai/blog/memory-fundamentals-a-model-for-agent-memory-engines</id>
    <published>2026-11-09T00:00:00.000Z</published>
    <updated>2026-11-09T00:00:00.000Z</updated>
    <author>
      <name>The ebbmem team at Spectrayan</name>
    </author>
    <category term="Research &amp; Foundations" />
    <summary>Why classifying agent memory by storage topology misses the point, and how the open Memory Fundamentals specification (MF-001) defines a recall algebra, normal forms, and conformance rules for memory engines.</summary>
  </entry>
  <entry>
    <title>Why Agent Memory Needs a Delete Button</title>
    <link href="https://ebbmem.ai/blog/why-agent-memory-needs-a-delete-button" rel="alternate" />
    <id>https://ebbmem.ai/blog/why-agent-memory-needs-a-delete-button</id>
    <published>2026-10-29T00:00:00.000Z</published>
    <updated>2026-10-29T00:00:00.000Z</updated>
    <author>
      <name>The ebbmem team at Spectrayan</name>
    </author>
    <category term="Data Privacy &amp; Architecture" />
    <summary>Why unbounded context accumulation creates data governance liabilities, and how active forgetting, byte zeroization, and cryptographic erasure receipts make long-term agent memory compliant and safe.</summary>
  </entry>
  <entry>
    <title>Memory Without an LLM in the Loop</title>
    <link href="https://ebbmem.ai/blog/memory-without-an-llm-in-the-loop" rel="alternate" />
    <id>https://ebbmem.ai/blog/memory-without-an-llm-in-the-loop</id>
    <published>2026-10-22T00:00:00.000Z</published>
    <updated>2026-10-22T00:00:00.000Z</updated>
    <author>
      <name>The ebbmem team at Spectrayan</name>
    </author>
    <category term="Architecture &amp; Infrastructure" />
    <summary>Eliminating external language models from the memory write path using local zero-shot ONNX models, FastEmbed, and deterministic civil date resolution to build sovereign, cost-predictable agent memory.</summary>
  </entry>
</feed>
