--- schema: 1 type: video title: >- Why your agents need decision traces, not just documents — Zach Blumenfeld, Neo4j description: >- A knowledge base tells a financial analyst agent the risk factors. A context graph tells it whether to reject or accept, because it also carries past decision traces, the reasoning behind them, and how similar cases resolved. Zach from Neo4j walks through how context graphs extend a standard RAG setup with three layers: short term conversation history, long term extracted entities, and reasoning traces that embed into vectors so structurally similar past decisions surface alongside semantically similar ones. The fastest path in is `uvx create-context-graph`, a one-command scaffold that gives you a backend, frontend, demo data, and an MCP server. It ships with 22 built-in domains or generates a graph ontology from a custom domain you describe. The underlying `neo4j-agent-memory` package handles entity extraction through a spaCy to GLiNER to LLM pipeline with deduplication and merging baked in, and plugs into pydantic AI, LangGraph, Crew, Google ADK, and others. Speaker info: - https://www.linkedin.com/in/zachblumenfeld/ --- Watch along with the corrected transcript, summary, timestamps and resources on this talk's official AIE page: https://ai.engineer/talks/B9h9ovW5H9U-why-your-agents-need-decision-traces-not-just-documents-zach-blumenfeld-neo4j Join our upcoming AI Engineer conferences (see https://ai.engineer landing page): - AIEi Shanghai, November 5–6 — the first for meeting top Chinese labs and engineers - AIE CODE in SF, November 10–12 — the top AI Coding conference returns - AIEi Sydney, December 7–8 — the first AIEi run alongside NeurIPS 2026 in Sydney --- canonical: >- https://liquidgold.cc/channel/aidotengineer/explore/video/why-your-agents-need-decision-traces-not-just-docum-37624d9d8060 markdown: >- https://liquidgold.cc/channel/aidotengineer/video/why-your-agents-need-decision-traces-not-just-docum-37624d9d8060.md llms: null parent: 'https://liquidgold.cc/channel/aidotengineer.md' channel: aidotengineer updated: '2026-10-24T10:18:07.537Z' language: null children: [] videoId: B9h9ovW5H9U durationSeconds: 1212 publishedAt: '2026-05-29T16:00:33.000Z' searchChunks: - ordinal: 0 text: null startSeconds: 25 endSeconds: 104 watch: 'https://liquidgold.cc/watch/B9h9ovW5H9U?t=25' heldReason: rights - ordinal: 1 text: null startSeconds: 104 endSeconds: 115 watch: 'https://liquidgold.cc/watch/B9h9ovW5H9U?t=104' heldReason: rights - ordinal: 2 text: null startSeconds: 115 endSeconds: 203 watch: 'https://liquidgold.cc/watch/B9h9ovW5H9U?t=115' heldReason: rights - ordinal: 3 text: null startSeconds: 115 endSeconds: 203 watch: 'https://liquidgold.cc/watch/B9h9ovW5H9U?t=115' heldReason: rights - ordinal: 4 text: null startSeconds: 214 endSeconds: 327 watch: 'https://liquidgold.cc/watch/B9h9ovW5H9U?t=214' heldReason: rights - ordinal: 5 text: null startSeconds: 374 endSeconds: 450 watch: 'https://liquidgold.cc/watch/B9h9ovW5H9U?t=374' heldReason: rights - ordinal: 6 text: null startSeconds: 450 endSeconds: 462 watch: 'https://liquidgold.cc/watch/B9h9ovW5H9U?t=450' heldReason: rights - ordinal: 7 text: null startSeconds: 450 endSeconds: 462 watch: 'https://liquidgold.cc/watch/B9h9ovW5H9U?t=450' heldReason: rights - ordinal: 8 text: null startSeconds: 462 endSeconds: 541 watch: 'https://liquidgold.cc/watch/B9h9ovW5H9U?t=462' heldReason: rights - ordinal: 9 text: null startSeconds: 541 endSeconds: 553 watch: 'https://liquidgold.cc/watch/B9h9ovW5H9U?t=541' heldReason: rights - ordinal: 10 text: null startSeconds: 553 endSeconds: 643 watch: 'https://liquidgold.cc/watch/B9h9ovW5H9U?t=553' heldReason: rights - ordinal: 11 text: null startSeconds: 553 endSeconds: 643 watch: 'https://liquidgold.cc/watch/B9h9ovW5H9U?t=553' heldReason: rights - ordinal: 12 text: null startSeconds: 653 endSeconds: 757 watch: 'https://liquidgold.cc/watch/B9h9ovW5H9U?t=653' heldReason: rights - ordinal: 13 text: null startSeconds: 653 endSeconds: 757 watch: 'https://liquidgold.cc/watch/B9h9ovW5H9U?t=653' heldReason: rights - ordinal: 14 text: null startSeconds: 653 endSeconds: 757 watch: 'https://liquidgold.cc/watch/B9h9ovW5H9U?t=653' heldReason: rights - ordinal: 15 text: null startSeconds: 768 endSeconds: 895 watch: 'https://liquidgold.cc/watch/B9h9ovW5H9U?t=768' heldReason: rights - ordinal: 16 text: null startSeconds: 768 endSeconds: 895 watch: 'https://liquidgold.cc/watch/B9h9ovW5H9U?t=768' heldReason: rights - ordinal: 17 text: null startSeconds: 768 endSeconds: 895 watch: 'https://liquidgold.cc/watch/B9h9ovW5H9U?t=768' heldReason: rights - ordinal: 18 text: null startSeconds: 986 endSeconds: 997 watch: 'https://liquidgold.cc/watch/B9h9ovW5H9U?t=986' heldReason: rights - ordinal: 19 text: null startSeconds: 997 endSeconds: 1087 watch: 'https://liquidgold.cc/watch/B9h9ovW5H9U?t=997' heldReason: rights - ordinal: 20 text: null startSeconds: 1087 endSeconds: 1099 watch: 'https://liquidgold.cc/watch/B9h9ovW5H9U?t=1087' heldReason: rights - ordinal: 21 text: null startSeconds: 1099 endSeconds: 1212 watch: 'https://liquidgold.cc/watch/B9h9ovW5H9U?t=1099' heldReason: rights --- # Why your agents need decision traces, not just documents — Zach Blumenfeld, Neo4j A knowledge base tells a financial analyst agent the risk factors. A context graph tells it whether to reject or accept, because it also carries past decision traces, the reasoning behind them, and how similar cases resolved. Zach from Neo4j walks through how context graphs extend a standard RAG setup with three layers: short term conversation history, long term extracted entities, and reasoning traces that embed into vectors so structurally similar past decisions surface alongside semantically similar ones. The fastest path in is \`uvx create-context-graph\`, a one-command scaffold that gives you a backend, frontend, demo data, and an MCP server. It ships with 22 built-in domains or generates a graph ontology from a custom domain you describe. The underlying \`neo4j-agent-memory\` package handles entity extraction through a spaCy to GLiNER to LLM pipeline with deduplication and merging baked in, and plugs into pydantic AI, LangGraph, Crew, Google ADK, and others. Speaker info: - https://www.linkedin.com/in/zachblumenfeld/ --- Watch along with the corrected transcript, summary, timestamps and resources on this talk's official AIE page: https://ai.engineer/talks/B9h9ovW5H9U-why-your-agents-need-decision-traces-not-just-documents-zach-blumenfeld-neo4j Join our upcoming AI Engineer conferences (see https://ai.engineer landing page): - AIEi Shanghai, November 5–6 — the first for meeting top Chinese labs and engineers - AIE CODE in SF, November 10–12 — the top AI Coding conference returns - AIEi Sydney, December 7–8 — the first AIEi run alongside NeurIPS 2026 in Sydney --- Canonical: https://liquidgold.cc/channel/aidotengineer/explore/video/why-your-agents-need-decision-traces-not-just-docum-37624d9d8060 · Channel: [AI Engineer](https://liquidgold.cc/channel/aidotengineer.md) > Search chunk text is held until public-source provenance is verified. ## Document chunks > Missing: no stored document chunks are available.