Slides: Connecting the Dots with Context Graphs — Stephen Chin, Neo4j
Source Video
Connecting the Dots with Context Graphs — Stephen Chin, Neo4j
Relationship To World's Fair 2026
These slides are extracted from a public AI Engineer YouTube video connected to World's Fair 2026. Speaker-matched clips are supporting context unless later confirmed as exact session recordings; official livestream recordings are day-level/event-level source material.
Related Scheduled Sessions
- No individual scheduled session mapping has been assigned yet; treat this as an event livestream deck.
Extracted Slides
OCR text:
PLATINUM SPONSORS
Braintrust WorkOS OpenAI
OCR text:
AlEngineer
Connecting thed
EUROPE
withcontextgrap
Stephen Chin(@steveonjava)
VPof Developer RelationsNeo4]
AlEngineer
OCR text:
AIE
AI Engineer
EUROPE
OCR text:
Scattered,lostreasoning Today
AIE CRM Why? Alternatives Considered?
Slack Slack ZendeskSupport Ticket
JraProject Board
pd
PagerDuty Zoom
Context Lose inPeople's Heads& Silos Tribal Knowledge,Cross-System Synthess ApprovalChains Outside Systems
AIEngineer
EUROPE
OCR text:
Hype cycle for Agentic Al
AIE EXPECTATIONS Hunan-Agent Collaborbonwospace AIAgentDe Domai Context Graphs Emtarprine Al Assistants- eticAIGo AgeticA/Security Agentic Analytics Multingeet Systes No-Code Agent Buders AlCodingAgonts Machine O Model Conlext Protocol certcAl stomers
EmbodedAl
AlAgnt Communicaticn Protoool Agent Managam tPatform
AlAgentsin SofthwareEngine Agent Marketpiace
FinOgs for Agentic Al AgenticBrowser Coetinul Leaig Agentc Commerce
Guardlan Agent Agent Experonce(Ax Computer Use for Agents Agent Deveiopent Lre Cycle Ageet Orchestration Platesu will be reached: 2ym 2-5y O5-10ym 10yrx Obsclete before plateau AsetAgri2026
Innovation Peak of Inflated Expectations Disillusionment Troughof TIME Enlightenment Productivity Plateauof
AIEngineer
EUROPE
OCR text:
AIE
AI Engineer
EUROPE
OCR text:
Knowledge graph components of
Narre “Dan” Cee
eS AIL) ene Y LS)
at N odes Arce cuane SC
* a Earns
represent entitics in the graph ie) we!
ecleniind
dl . A & be ean &
Relationships
G a
a stop echt tet lene Reg an ~~
interactions between nodes aa 7
Jan 10, 2011
i
o=0
Properties
; 7 ory
ice] Ike ei pL area GAL DIU Ce MON MR IOIG (3101) Hae oh
relationships including vectors, can be indexed Bheccen pt oo “An executive car manutactured and...”
eae eae (ORC OR SOO |
eR RMERMaS TC UT SI SaRee ai seee eat Serr SE CT 22
i
A
Al Engineer
EUROPE
OCR text:
Enhance relevance with domain context a
What was in the Care Plans associated with Andrea Jenkins'’s emphysema?
Pane a
ba bd
“ae
bd ad
* vs bd
LLM Direct
TTT over Neodj ay call
eet ae wy i wot
connected
ic a OP ee C toreal
ee oo
| Al Engineer |
EUROPE
OCR text:
. A a)
Enhance relevance with domain context ;
What was in the Care Plans associated with Andrea Jenkins'’s emphysema?
ae Baseline RAG
. ‘ OEP hese VCR stats ey mel od Me Ae)
* re and Chroma vector database
Grounded but
Incomplete
| Al Engineer |
EUROPE
OCR text:
Enhance relevance with domain context a.
What was in the Care Plans associated with Andrea Jenkins'’s emphysema?
Pane a
Pd a
ae
* 4
bd * 5d
GraphRAG
Neo4j, LangChain,
pres ee ree
fotevantolintten i sea & 7 - ' - .
a heen, errno
| Al Engineer |
aU elas
OCR text:
I KNOW KUNG FU
Engineering the future of AI
OCR text:
AA
Long-term memory 7
Persistent knowledge graph of entities, relationships, and learned preferences
rane a
* bd
aco
va ia a Person eae ne Te (ot eat Core Rene banana Saleh Canes Cad el ae ona ed
a *
: (Oleecein 1 aeh tole Meee cerca en (eee eT Ser CTS ore ee COOL LS
eee ee one nth cea te eC EeaReel ee Ta
Tre ble ate eet Ed Cr cd
Aco te pebae pe erent Race Oia eee ee eS
Google DeepMind
OCR text:
Wal
Reasoning memory ;
Decision traces, tool usage audits, and provenance — the layer that makes Al explainable
a * &
. * Tool cell treices Decision provenance
ra Pa Eseory toolopyccat un, | V1 Why cad the geet ohcose Us path?
ara Paramerers Pew Ges Cavsal chain tubs recerdecs
eecrpiete aud t tra |
Learning from Complicnce &
experience 3 4 debugging
Agentebeces ditscwed SOs MCS CELE Sie St
potetiine sep lar before Wore. youd cae Peace pectly
Telos sl ceeofu palteres wat nappened ard why
| Al Engineer |
SUL ela
OCR text:
Building memory
Tool call traces
1
Learning from experience
3
OCR text:
a)
Why graphs for agent memory? 7
Pond aa Q
© il Connec tons betwcer entities. decisions, and Foilaw chains: Customer + Account +
bad * events are the data potan atterthougnt Trapsaction + Deoson +» Polcy + Employer
Sg ad
areas
o @
Graph embedd nes (PasthRPi tind s.milar Ful provenance chain trace exactly
situations by 1eteors topoiosy, mot rust text how and why every decis on was made
(@) co
& i
roams
Information lesrned in ane conversanan Necdy ACID compliance, enteror se scale.
PWR CTerirerol Calla ma) tne Lee cutan Cera Ten Orel ars) proven technolosy
Engineering the fut A
OCR text:
. a
Neo4j-agent memory :
Complete Agent Memory API
Po : ; ‘
Short-Term Long-Term Reasoning
Bd 4 Cometsater cistern & Knowledge grach of entities Context graphs for
we va Picoen Meraal Coral of relator ships Pore Sa nnn Dm Laas Cord aT roe
rd
add_message() add_entity() start_trace()
PT rice) Cele) TF ied ult Sst TO) irate mest sO)
oI T ES Cle! TSO eee eto) record_tool_call()
OATS e TT TO} get_locations_near() rors OS ee ale Tee)
Cote T rt e] PLN SS Se eT TO) Pte ree Teh rs
'
Neo4j Context Graph
Vector Search + Graph Traversal
| Al Engineer |
SUL ela 3
OCR text:
“ a eR keuperrony verce app e > mg 1 7 x
~ DP Conversations ZG) Lenny’s Memory O Abou Po Gua
° ©
Cd Agent Configurapen
Don
° & Ayere Copatete et
- —— , Ask about Lenny’s Podcast
Pel 2” Bar ep teas coe:
° Use memory graph search to bet Pte wart wR we : 5
A explore what quests ay about aan wiee ee dwn 2 oe unm Be
re r 9 erameteeonnny
. Shon me Incotions Tennaned © “fF Faplore a Top< “Top Companes ” Se Te
ty * . the Bean Cesena 2 " sr Who ts Bran Pa .
. oo oo . : . “8 figieee tan aiey
Use memary geaph search 10 woe
fexgilure ate! guests uy about
a Parris +: Speaker Quotes oh Map Wiew Seema miige La aes
how do Dat conednste 4 4 . ao “ :
. . * we wae i P Restate locas a
OES Ee Xo lead Call Cart ?
py Eheotpabuagent. 2,
| HISD we
| Al Engineer |
OCR text:
Contextgraphbasedretrieval
AIE Search ContextGraphMemory RetrievalTools AgentLoop Contextual Memory
Retrieval Query Information Relevant
neo4j
Knowledge Graph + Vectors GraphData Science
Engineering the future of Al
OCR text:
£2 A 5 al
Context graph: financial services data model ;
Financial Services Context Graph
Ss ENTITIES (What Exists)
id * SCH aot me Cetera eA a PTicea UST le-esnPen aera)
au Cd ie ee
PY oy
ket :
EVENTS (What Happened)
Decis:ons, Transactions, Approvals, Rejections =
CONTEXT (The Why) oo nel me
Policies Applied. Risk Factors. Employee Reasoning
De
oer
Engineering the future of Al
OCR text:
Context graph demo architecture
Ai-powered decision tracing for financial institutions
DATA SOURCES
Support Ticket System
CRM (e.g. Salesforce)
Internal Business Data
Engineering the future of AI
OCR text:
AIEngineer
EUROPE
LUIACC
2026 AlEngineer
OCR text:
AIE
Engineering the future of AI
OCR text:
a)
Neodgj
Graph Academy
re rf
7 rd 7 EAS aaa Sree eee ceca Cee SeaTac
RS Re eee SOOM ere ecm eM eer OCR AM] me ON ee Rea Bese
WE heer tes Larceny obey
Oe
See
https://dev.neo4j.com/ga-rag
Google DeepMind
Slide-Derived Subjects To Review
Subject extraction uses video title, related session titles/descriptions, transcript context, and OCR text when available. OCR is best-effort and should be reviewed against the embedded slide images.