---
title: "Building the Document Context Layer for AI Agents"
category: "talks"
date: "2026-06-29"
time: "11:10am-11:30am"
track: "Vision & OCR"
room: "Track 2"
speakers: ["Jerry Liu"]
sourceLabels: ["Official conference schedule", "Public YouTube metadata"]
scheduleTrack: "Vision & OCR"
scheduleRoom: "Track 2"
scheduleLabels: ["Vision & OCR", "Track 2", "sponsor", "confirmed"]
---
# Building the Document Context Layer for AI Agents

## Conference Context
- Date/time: 2026-06-29 · 11:10am-11:30am
- Track/room: Vision & OCR · Track 2
- Speaker(s): Jerry Liu
- Session type/status: sponsor · confirmed

- Track: Vision & OCR
- Room: Track 2
- Session type: sponsor
- Status: confirmed

## Session Description
AI agents are the new knowledge workers, but knowledge work depends on unstructured enterprise context. ~90% of that data lives in the form of document containers - from human-native (PDFs, Word, Pptx) to emerging agent-native formats (HTML, MD). Doing RAG in 2026 involves generalized agent harnesses with tools, MCPs, and skills. In this world, every company building agents needs a Document Context Layer, the bridge between their unstructured docs and the agents trying to reason over them. This talk covers what that layer looks like in practice: from document understanding, retrieval, and workflows, to areas yet to be explored — agent-native formats, versioning, editing, permissions, and longer-running agents.

## Media Evidence
[Building AI Agents that actually automate Knowledge Work - Jerry Liu, LlamaIndex](https://www.youtube.com/watch?v=jVGCulhBRZI) (speaker-match related prior/adjacent AI Engineer video; captions: English auto-captions).

- Source video: `youtube-jVGCulhBRZI`
- Slide deck: [[youtube-jVGCulhBRZI-dense-slides|Dense Slides: Building AI Agents that actually automate Knowledge Work - Jerry Liu, LlamaIndex]] — 4 visible slide image(s); 4 HTML recreation(s).
![[assets/dense-slides/jVGCulhBRZI/slide-001.jpg]]
![[assets/dense-slides/jVGCulhBRZI/slide-002.jpg]]
![[assets/dense-slides/jVGCulhBRZI/slide-003.jpg]]
- Additional slide evidence: [[youtube-jVGCulhBRZI-slides|Slides: Building AI Agents that actually automate Knowledge Work - Jerry Liu, LlamaIndex]], [[youtube-jVGCulhBRZI-reconstructed-slides|Reconstructed Slides: Building AI Agents that actually automate Knowledge Work - Jerry Liu, LlamaIndex]]
- Slide-derived themes for `youtube-jVGCulhBRZI`: data, knowledge, unstructured, documents, automate, better, talking, does.

## Evidence Graph
This evidence graph is generated from currently linked source material: official schedule text, related video pages, cached transcripts, visible slide text, dense/reconstructed slide pages, and AI slide-classification audits.

### Media Signals
- `youtube-jVGCulhBRZI` — 9 slide-derived text signals
- Slide-derived themes for `youtube-jVGCulhBRZI`: data, knowledge, unstructured, documents, automate, better, talking, does.
- Evidence links for `youtube-jVGCulhBRZI`: [[youtube-jVGCulhBRZI]], [[youtube-jVGCulhBRZI-slides]], [[youtube-jVGCulhBRZI-dense-slides]], [[youtube-jVGCulhBRZI-reconstructed-slides]]

### Agent Reading Notes
Use these signals to refine the synopsis, topic links, people/company context, and method notes. If a source is a related external video rather than an exact official recording, keep it framed as supporting evidence.

## Transcript Status
Related video transcript availability: English auto-captions. Treat this as supporting context, not a recording of this exact scheduled session unless later confirmed. Not fetched yet.

## People
- [[jerry-liu]]

## Supporting Slides
- [[youtube-jVGCulhBRZI-slides]] — extracted from the related public AI Engineer video.

## Slide Evidence
- Slide-only cropped deck: [[youtube-jVGCulhBRZI-dense-slides]] (4 viable slide images).
- Related slide/OCR pages:
- [[youtube-jVGCulhBRZI-dense-slides]]
- [[youtube-jVGCulhBRZI-reconstructed-slides]]
- [[youtube-jVGCulhBRZI-slides]]
- Slide-derived terms: `automation`, `knowledge`, `work`, `microsoft`, `cases`, `time`, `ieee`, `unstructured`, `documents`, `excel`, `agentic`, `financial`, `extraction`, `document`, `awws`, `graphite`, `windsurf`, `moneobp`

## Synthesis
### Synthesized Breakdown
# Building the Document Context Layer for AI Agents ## Conference Context - Date/time: 2026-06-29 · 11:10am-11:30am - Track/room: Vision & OCR · Track 2 - Speaker(s): Jerry Liu - Session type/status: sponsor · confirmed - Track: Vision & OCR - Room: Track 2 - Session type: sponsor - Status: confirmed ## Session Description AI agents are the new knowledge workers, but knowledge work depends on unstructured enterprise context. ~90% of that data lives in the form of document containers - from human-native (PDFs, Word, Pptx) to emerging agent-native formats (HTML, MD). Doing RAG in 2026 involves generalized agent harnesses with tools, MCPs, and skills. In this world, every company building agents needs a Document Context Layer, the bridge between their unstructured docs and the agents trying to reason over them.

### Speaker And Company Context
- [[jerry-liu|Jerry Liu]] — CEO at [[llamaindex|LlamaIndex]].

### Topics Covered
- [[agent-security]]
- [[agentic-search]]
- [[agentic-web]]
- [[mcp]]

### Derived Links And Source Material
- [[youtube-jVGCulhBRZI]] — related YouTube source page.
- [[youtube-jVGCulhBRZI-slides]] — slide evidence.
- [[youtube-jVGCulhBRZI-reconstructed-slides]] — slide evidence.
- [[youtube-jVGCulhBRZI-dense-slides]] — slide evidence.

### Novel Concepts / Clever Methods
- No highlighted novel concept has been detected yet.

### Evidence Boundary
This synthesis is based on the official schedule and linked source pages. It should be revisited when exact session recordings or transcript-backed secondary sources are available.