vidtheque.

read-only demo

Videos abvQEhvRI_c

Context Graphs for Explainable, Decision-Aware AI Agents — Andreas Kollegger & Zaid Zaim, Neo4j

index_state ready data_status ok

AI Engineer· published 2026-05-28· 0:16:38· en-US· indexed 2026-08-10 19:56

Open on YouTube

Scene timeline

  1. Shot 0, 0:00 to 0:05, 1 of 1 keyframes kept
  2. Shot 1, 0:05 to 0:09, 1 of 1 keyframes kept
  3. Shot 2, 0:09 to 0:14, 1 of 1 keyframes kept
  4. Shot 3, 0:14 to 0:32, 1 of 1 keyframes kept
  5. Shot 4, 0:32 to 0:38, 1 of 1 keyframes kept
  6. Shot 5, 0:38 to 1:26, 1 of 1 keyframes kept
  7. Shot 6, 1:26 to 1:27, 1 of 1 keyframes kept
  8. Shot 7, 1:27 to 1:28, 0 of 1 keyframes kept
  9. Shot 8, 1:28 to 1:58, 0 of 1 keyframes kept
  10. Shot 9, 1:58 to 2:28, 0 of 1 keyframes kept
  11. Shot 10, 2:28 to 2:37, 1 of 1 keyframes kept
  12. Shot 11, 2:37 to 2:58, 1 of 1 keyframes kept
  13. Shot 12, 2:58 to 3:34, 1 of 1 keyframes kept
  14. Shot 13, 3:34 to 3:54, 1 of 1 keyframes kept
  15. Shot 14, 3:54 to 4:12, 0 of 1 keyframes kept
  16. Shot 15, 4:12 to 4:14, 1 of 1 keyframes kept
  17. Shot 16, 4:14 to 4:40, 1 of 1 keyframes kept
  18. Shot 17, 4:40 to 5:05, 1 of 1 keyframes kept
  19. Shot 18, 5:05 to 5:31, 1 of 1 keyframes kept
  20. Shot 19, 5:31 to 5:54, 0 of 1 keyframes kept
  21. Shot 20, 5:54 to 5:56, 1 of 1 keyframes kept
  22. Shot 21, 5:56 to 6:21, 1 of 1 keyframes kept
  23. Shot 22, 6:21 to 6:47, 0 of 1 keyframes kept
  24. Shot 23, 6:47 to 6:49, 1 of 1 keyframes kept
  25. Shot 24, 6:49 to 7:08, 1 of 1 keyframes kept
  26. Shot 25, 7:08 to 7:54, 1 of 1 keyframes kept
  27. Shot 26, 7:54 to 8:21, 0 of 1 keyframes kept
  28. Shot 27, 8:21 to 8:27, 1 of 1 keyframes kept
  29. Shot 28, 8:27 to 8:47, 1 of 1 keyframes kept
  30. Shot 29, 8:47 to 9:08, 1 of 1 keyframes kept
  31. Shot 30, 9:08 to 9:30, 1 of 1 keyframes kept
  32. Shot 31, 9:30 to 9:48, 1 of 1 keyframes kept
  33. Shot 32, 9:48 to 10:14, 1 of 1 keyframes kept
  34. Shot 33, 10:14 to 10:49, 1 of 1 keyframes kept
  35. Shot 34, 10:49 to 11:24, 0 of 1 keyframes kept
  36. Shot 35, 11:24 to 11:55, 1 of 1 keyframes kept
  37. Shot 36, 11:55 to 12:26, 1 of 1 keyframes kept
  38. Shot 37, 12:26 to 12:57, 0 of 1 keyframes kept
  39. Shot 38, 12:57 to 13:31, 0 of 1 keyframes kept
  40. Shot 39, 13:31 to 14:11, 1 of 1 keyframes kept
  41. Shot 40, 14:11 to 14:50, 1 of 1 keyframes kept
  42. Shot 41, 14:50 to 15:28, 0 of 1 keyframes kept
  43. Shot 42, 15:28 to 16:02, 1 of 1 keyframes kept
  44. Shot 43, 16:02 to 16:12, 1 of 1 keyframes kept
  45. Shot 44, 16:12 to 16:23, 1 of 1 keyframes kept
  46. Shot 45, 16:23 to 16:38, 1 of 1 keyframes kept

46 shot(s).

keyframes kept every frame deduplicated

What was stored

cues
176
whisperx 176
chunks
29
from 176 cues
keyframes
35
kept of 46 captured
frames with text
35
732 lines read
chapters
0
from the source metadata
keyframe bytes
6.1 MB
word timings on 176 cues

Provenance

Each pipeline stage, its state and the model that produced it
stage state model started took
fetch done — 2026-08-10 19:10 1m 37s
stt done — 2026-08-10 19:11 17s
chunk done — 2026-08-10 19:12 0s
text_embed done — 2026-08-10 19:56 1s
keyframe done — 2026-08-10 19:12 1m 27s
ocr done — 2026-08-10 19:13 15s
frame_embed done — 2026-08-10 19:56 6s

Frames, and what the machine read

  • 0:03 #0 done2 line(s)

    shot 0·sharpness 658.4

    1. AlEngineer0.98
    2. EUROPE1.00

  • 0:08 #1 done2 line(s)

    shot 1·sharpness 827.9

    1. PRESENTINGSPONSOR1.00
    2. Google DeepMind1.00

  • 0:13 #2 done3 line(s)

    shot 2·sharpness 907.7

    1. PLATINUM SPONSORS0.98
    2. # Braintrust0.96
    3. WorkOS OpenAI0.95

  • 0:21 #3 done19 line(s)

    shot 3·sharpness 596.4

    1. AlEngineer0.99
    2. EUROPE1.00
    3. AlEngineer0.99
    4. APRIL 9, 2026 - 2.50PM0.97
    5. EUROPE1.00
    6. Context Graphs for1.00
    7. Explainable, Decision-1.00
    8. Al Agents0.98
    9. EU/ACC1.00
    10. Zaid Zaim0.92
    11. Andreas Kollegger0.98
    12. Developer Advocate0.98
    13. Director.0.99
    14. Neo4j0.97
    15. Developer Advocat0.99
    16. ;neo4j0.77
    17. Neo4j0.97
    18. AlEngineer0.98
    19. EUROPE1.00

  • 0:34 #4 done21 line(s)

    shot 4·sharpness 2299.2

    1. neo4j0.98
    2. AlEngineer0.99
    3. EUROPE1.00
    4. APRIL 9, 2026 - 2.50PM0.96
    5. Context Graphs for1.00
    6. AIE1.00
    7. ★1.00
    8. Explainable, Decision-Aware1.00
    9. ★1.00
    10. Al Agents0.97
    11. Zaid Zaim1.00
    12. Andreas Kollegger1.00
    13. Developer Advocate1.00
    14. Director,1.00
    15. Neo4j1.00
    16. Developer Advocate1.00
    17. Neo4j1.00
    18. :neo4j0.93
    19. AlEngineer0.96
    20. AlEngineer1.00
    21. EUROPE1.00

  • 1:02 #5 done20 line(s)

    shot 5·sharpness 1538.2

    1. Knowledge Graphs Unlock GenAI0.98
    2. Knowledge1.00
    3. Language1.00
    4. AIE1.00
    5. ★1.00
    6. Context1.00
    7. KGs1.00
    8. LLMs1.00
    9. Reasoning1.00
    10. ★1.00
    11. ★1.00
    12. ★1.00
    13. Enrichment1.00
    14. Creativity1.00
    15. BETTERTOGETHER1.00
    16. neo4j0.93
    17. neo4j0.76
    18. AlEngineer0.97
    19. AlEngineer1.00
    20. EUROPE1.00

  • 1:27 #6 done42 line(s)

    shot 6·sharpness 3097.9

    1. neo4j-agent-memory1.00
    2. Complete Agent Memory API0.99
    3. Short-Term0.99
    4. Long-Term1.00
    5. Reasoning1.00
    6. ★1.00
    7. ★1.00
    8. Conversation history0.98
    9. Knowledge graph of0.98
    10. Context graphs for0.99
    11. ★1.00
    12. AIE1.00
    13. & session context1.00
    14. entities & relationships0.97
    15. explainability & learning1.00
    16. ★1.00
    17. ★1.00
    18. ★1.00
    19. ★1.00
    20. ★1.00
    21. add_message()1.00
    22. add_entity()1.00
    23. start_trace()1.00
    24. search_messages()0.97
    25. search_entities()0.98
    26. record_step()1.00
    27. get_session()1.00
    28. add_preference()1.00
    29. record_tool_call()1.00
    30. get_recent_messages()1.00
    31. search_locations_near()1.00
    32. get_similar_traces()1.00
    33. delete_session()1.00
    34. get_entity_graph()1.00
    35. get_trace_provenance()1.00
    36. Neo4j Context Graph0.98
    37. Vector Search + Graph Traversal0.99
    38. Context Graph Meetup SF · Neo4j0.98
    39. ;neo4j0.80
    40. AlEngineer0.96
    41. AlEngineer0.99
    42. EUROPE1.00

  • 1:28 #7 skipped

    shot 7·duplicate of #5

  • 1:41 #8 skipped

    shot 8·duplicate of #6

  • 2:25 #9 skipped

    shot 9·duplicate of #6

  • 2:29 #10 done38 line(s)

    shot 10·sharpness 4627.5

    1. Neo4i Graph Components0.99
    2. Nodes represent entities in the1.00
    3. Name: “Dan"0.95
    4. Name: “Ann"0.95
    5. graph1.00
    6. Born: May 29, 19701.00
    7. Twitter: "@dan"0.98
    8. Born: Dec 5, 19750.98
    9. KNOWS1.00
    10. KNOWS1.00
    11. AIE1.00
    12. ★1.00
    13. Relationships represent1.00
    14. Person1.00
    15. Person1.00
    16. ★1.00
    17. associations or interactions between1.00
    18. LIVES WITH1.00
    19. ★1.00
    20. ★1.00
    21. ★1.00
    22. nodes1.00
    23. DRIVES1.00
    24. Since:1.00
    25. OWNS1.00
    26. Jan 10, 20110.98
    27. Properties represent attributes of0.99
    28. nodes or relationships including1.00
    29. Car1.00
    30. vectors, can be indexed1.00
    31. Brand "Volvo"0.95
    32. Model: "V70"0.96
    33. Description: "An executive car manufactured and..."1.00
    34. DescEmbedding: [0.1, -0.3, 0.4, .., -0.7]0.97
    35. DescSource:"https://en.wikipedia.org/wiki/Volvo_V70"1.00
    36. ;neo4j0.79
    37. Engineering the future of Al0.99
    38. AlEngineer1.00

  • 2:52 #11 done16 line(s)

    shot 11·sharpness 504.4

    1. Graph Compone0.99
    2. Noo0.89
    3. sent entities in the1.00
    4. AlEngineer1.00
    5. graph1.00
    6. EUROPE1.00
    7. Relationships represent1.00
    8. associations or interactions between1.00
    9. nodes1.00
    10. EU/ACC0.94
    11. Properties represent attributes of1.00
    12. nodes or relationships including0.98
    13. vectors, can be indexed0.99
    14. :neo4j0.86
    15. AlEngineer0.97
    16. EUROPE1.00

  • 3:16 #12 done25 line(s)

    shot 12·sharpness 3188.2

    1. What is a Context Graph?0.98
    2. A knowledge graph specifically designed to capture decision traces — the full context, reasoning, and1.00
    3. causal relationships behind every significant decision.0.99
    4. AIE1.00
    5. ★1.00
    6. ★1.00
    7. Traditional Audit Log1.00
    8. Context Graph0.97
    9. ★1.00
    10. Records actions only1.00
    11. Captures the full "why"0.99
    12. ★1.00
    13. "Transaction rejected at 14:32"1.00
    14. Decision traces & causal chains1.00
    15. No relationships between events1.00
    16. Entities, relationships, events0.99
    17. No causal chain or reasoning1.00
    18. Tribal knowledge made queryable0.99
    19. Flat, disconnected records1.00
    20. Connected, traversable structure0.98
    21. Jaya Gupta & Ashu Garg, "Al's Trillion-Dollar Opportunity: Context Graphs" — Foundation Capital, Dec 20250.99
    22. :neo4j0.81
    23. AlEngineer0.95
    24. AlEngineer1.00
    25. EUROPE1.00

  • 3:42 #13 done24 line(s)

    shot 13·sharpness 3062.9

    1. The Missing "Why"0.98
    2. When an Al agent recommends approving a $100K credit line increase,0.98
    3. the question isn't just "is this the right answer?"1.00
    4. ***0.57
    5. ★1.00
    6. AIE1.00
    7. ★1.00
    8. ★1.00
    9. ★1.00
    10. No Memory0.96
    11. No Audit Trail0.99
    12. No Shared Learning1.00
    13. When something goes wrong,1.00
    14. Ask an agent about last week's1.00
    15. Deploy multiple agents and they0.99
    16. conversation — blank stare1.00
    17. no one knows why the agent1.00
    18. can't share what they've learned1.00
    19. decided that1.00
    20. 95% of Al pilots fail to deliver returns — MIT0.98
    21. neo40.86
    22. Engineering the future of Al1.00
    23. AlEngineer0.95
    24. 20260.98

  • 4:03 #14 skipped

    shot 14·duplicate of #11

  • 4:14 #15 done25 line(s)

    shot 15·sharpness 330.2

    1. is Are0.92
    2. AlEngineer0.99
    3. 阳0.62
    4. EUROPE1.00
    5. Suppliers1.00
    6. Route planning & optimizatic0.98
    7. Real-time shipment trackin0.98
    8. Inventory planning1.00
    9. Risk analysis1.00
    10. EU/ACC0.99
    11. Employees1.00
    12. Talent learning & development0.95
    13. Career management0.99
    14. Skills management0.97
    15. Orgs (who is wha)0.97
    16. Job search1.00
    17. Network & Security1.00
    18. Account/identity control0.99
    19. Reputation scoring0.96
    20. Threat detection1.00
    21. :neo4j0.87
    22. Access control1.00
    23. Zero trust1.00
    24. AlEngineer1.00
    25. EUROPE1.00

  • 4:17 #16 done49 line(s)

    shot 16·sharpness 3290.8

    1. Graphs Are Everywhere1.00
    2. Customers1.00
    3. Recommendations1.00
    4. 回0.64
    5. Loyalty programs1.00
    6. Churn prevention1.00
    7. Customer offers1.00
    8. Dynamic pricing1.00
    9. AIE1.00
    10. Product1.00
    11. Intelligent ads1.00
    12. ★1.00
    13. ★1.00
    14. ★1.00
    15. ★1.00
    16. Route planning & optimization0.98
    17. Real-time shipment tracking1.00
    18. Suppliers1.00
    19. New product introductions0.99
    20. Product recommendations1.00
    21. Product customizations0.98
    22. Inventory planning0.99
    23. Product inventory1.00
    24. Risk analysis1.00
    25. Product pricing1.00
    26. Employees1.00
    27. Talent learning & development0.99
    28. Career management1.00
    29. Anti-money laundering1.00
    30. Finance1.00
    31. Skills management1.00
    32. Network & Security1.00
    33. Circular payments0.97
    34. Orgs (who is who)1.00
    35. Fraud detection1.00
    36. Job search1.00
    37. Account/identity control1.00
    38. Reputation scoring1.00
    39. Process1.00
    40. Threat detection0.97
    41. Bottleneck identification1.00
    42. Access control1.00
    43. Process improvement1.00
    44. Zero trust0.99
    45. Process automation1.00
    46. Process monitoring1.00
    47. neo4j0.88
    48. Engineering the future of Al1.00
    49. AlEngineer1.00

  • 5:02 #17 done49 line(s)

    shot 17·sharpness 3274.2

    1. Graphs Are Everywhere1.00
    2. Customers1.00
    3. Recommendations1.00
    4. 回0.66
    5. Loyalty programs1.00
    6. Churn prevention1.00
    7. Customer offers1.00
    8. Dynamic pricing1.00
    9. AIE1.00
    10. Product1.00
    11. Intelligent ads1.00
    12. ★1.00
    13. Suppliers1.00
    14. Product recommendations1.00
    15. ★1.00
    16. ★1.00
    17. Route planning & optimization0.98
    18. Real-time shipment tracking1.00
    19. New product introductions1.00
    20. Product customizations1.00
    21. Inventory planning1.00
    22. Risk analysis1.00
    23. Product inventory1.00
    24. Product pricing1.00
    25. Employees1.00
    26. Talent learning & development0.99
    27. Career management1.00
    28. Anti-money laundering1.00
    29. Finance1.00
    30. Skills management1.00
    31. Network & Security1.00
    32. Circular payments0.99
    33. Orgs (who is who)1.00
    34. Fraud detection1.00
    35. Job search1.00
    36. Account/identity control1.00
    37. Reputation scoring1.00
    38. Process1.00
    39. Threat detection0.98
    40. Bottleneck identification1.00
    41. Access control1.00
    42. Process improvement1.00
    43. Zero trust1.00
    44. Process automation1.00
    45. Process monitoring1.00
    46. ;neo4j0.72
    47. AlEngineer0.97
    48. AlEngineer1.00
    49. EUROPE1.00

  • 5:28 #18 done40 line(s)

    shot 18·sharpness 2525.5

    1. Memory Graph1.00
    2. Long-Term Memory0.98
    3. Legend1.00
    4. Short-Tern0.95
    5. Long-Tern0.90
    6. EntityPerson1.00
    7. Reasoning1.00
    8. MENTIONED_IN0.99
    9. **0.92
    10. Short-Term Memory0.99
    11. WORKS_AT0.99
    12. AIE1.00
    13. ★1.00
    14. ★1.00
    15. ★1.00
    16. ★1.00
    17. ★1.00
    18. Conversation1.00
    19. FIRST_MESSAGE0.96
    20. Message1.00
    21. Message0.98
    22. Message1.00
    23. Entity.Organization0.97
    24. NEXT_MESSAGE1.00
    25. NEXT_MESSAGE0.99
    26. TRIGGERED1.00
    27. RETRIEVED1.00
    28. HAS_STEP1.00
    29. USED_TOOL0.99
    30. ReasoningTrace1.00
    31. ReasoningStep0.99
    32. Toolcall1.00
    33. CALL_OF0.96
    34. Tool1.00
    35. 81.00
    36. Reasoning Memory1.00
    37. neo40.83
    38. AlEngineer0.96
    39. AlEngineer1.00
    40. EUROPE0.96

  • 5:47 #19 skipped

    shot 19·duplicate of #11

  • 5:55 #20 done17 line(s)

    shot 20·sharpness 570.3

    1. Agentic Gr0.98
    2. Agents1.00
    3. Graph1.00
    4. AlEngineer1.00
    5. Query1.00
    6. Retrieval Agent1.00
    7. EUROPE1.00
    8. LLM1.00
    9. (Planning)1.00
    10. User1.00
    11. EU/ACO0.90
    12. Response1.00
    13. LLM1.00
    14. (Output)1.00
    15. :neo4j0.94
    16. AlEngineer1.00
    17. EUROPE1.00

  • 5:59 #21 done38 line(s)

    shot 21·sharpness 3271.2

    1. Agentic GraphRAG1.00
    2. Users1.00
    3. Agents1.00
    4. Tools1.00
    5. Graph1.00
    6. Retrieval Agent1.00
    7. 3a1.00
    8. Vector + Graph1.00
    9. Query1.00
    10. Retriever1.00
    11. ***0.57
    12. AIE1.00
    13. 31.00
    14. 3b1.00
    15. ★1.00
    16. Text2Cypher1.00
    17. ★1.00
    18. ★1.00
    19. ★1.00
    20. Retriever1.00
    21. LLM1.00
    22. 3n1.00
    23. (Planning)1.00
    24. User1.00
    25. SME Cypher1.00
    26. Graph1.00
    27. Template1.00
    28. Database1.00
    29. Response1.00
    30. LLM1.00
    31. HRIS1.00
    32. MCP1.00
    33. HRIS1.00
    34. (Output)1.00
    35. Retriever1.00
    36. (neo4j0.75
    37. Engineering the future of Al1.00
    38. AlEngineer1.00

  • 6:29 #22 skipped

    shot 22·duplicate of #21

  • 6:48 #23 done8 line(s)

    shot 23·sharpness 1011.5

    1. AIE1.00
    2. ★1.00
    3. ★1.00
    4. Why care about decision making?1.00
    5. 10 Neo4j Inc. All rights reserved 20261.00
    6. Engineering the future of Al0.99
    7. AlEngineer1.00
    8. 20260.97

Transcript

176 cues· 2,790 words· 15,225 chars

  1. 0:14 Hello, good evening, developers.
  2. 0:18 How is everyone doing?
  3. 0:20 Welcome to iEngineer and welcome to the Neo4j Context Graph session number two.
  4. 0:26 Excited to be here together with my peer, ABK, to share with you more about context graphs, how they can help Asians to become more decision aware.
  5. 0:38 So one of the big challenges that we are trying to solve these days in the AI era is basically using knowledge graphs to unlock AI, where we want to use graphs to fill the gap of knowledge, give AI agents the right tools, and enrich them with the right content to do.
  6. 1:05 to solve certain challenges.
  7. 1:10 So we all know that AI agents are really good in language, reasoning, and creativity, but we want to miss or fill the missing puzzle piece of knowledge for the agents.
  8. 1:30 So, and big topic we have been, we also saw before at Steve's talk is memory.
  9. 1:36 So to, we talk about different types of memory, so short-term, long-term, and reasoning memory.
  10. 1:45 So in a short-term memory, we try to understand and capture the conversations that a user had with the AI agent.
  11. 1:57 And in a long-term memory, we capture more
  12. 2:03 contextual knowledge in this case, things around organizations, people and things basically.
  13. 2:11 So that really give us more of the generalized overview of the context.
  14. 2:18 And of course we need here reasoning, helping basically the AI to make the right decisions to achieve a certain task.
  15. 2:30 We saw also that, for the background of this, we need a graph.
  16. 2:39 So a graph basically is made out of nodes and relationships to basically help you understand the deep and complex relations and connections between different type of data.
  17. 2:55 So the big question is now why context graphs?
  18. 2:57 So we also have been talking also lately a lot about context engineering and now basically the big ask again why context graphs?
  19. 3:08 So basically we are still within the sphere of context engineering when we talk about context graph but context graph is really,
  20. 3:15 the shift of having um ai agents that already are very good in in providing knowledge um to uh to their users having the right tools and content and context but with context graphs we would really like to provide additionally to the knowledge the right rules
  21. 3:37 and policies to the agents to help basically them become more capable to drive decisions.
  22. 3:48 So not only knowledge but also decisions.
  23. 3:52 So that's why we're moving not only how an agent or what an agent can do but
  24. 3:57 Also now we want to really capture within a context graph the missing why.
  25. 4:01 So why an agent needs to do something, and this is something we capture with data that focuses on policies and rules.
  26. 4:13 So when we talk about Neo4j, we say also that graphs are everywhere.
  27. 4:20 So this is a basic replica of an organization.
  28. 4:25 Each organization is basically also reflected here from finance, or department of an organization, from finance to product to suppliers.
  29. 4:34 and so on and so forth.
  30. 4:34 We saw also a good example by Steve before on financial services.
  31. 4:39 So if someone is eligible to get a certain amount of money here, so yes or no.
  32. 4:47 So part of this chapter we are really
  33. 4:51 Going deeper with our customers and verticals to understand their needs and basically how we can help them solve the why question and help them drive better agents that can be more decision-aware.
  34. 5:08 This is also how a memory graph would look like.
  35. 5:13 So if we want to
  36. 5:15 to an agent to also drive the right decisions, we need to also give them memory capabilities.
  37. 5:23 So this is a quick overview of how the memory graph would look like.
  38. 5:27 In green we see the short-term memory.
  39. 5:29 So we are capturing again the conversations, the state history.
  40. 5:34 And then in long-term memory is more like the generalized overview of organizations, people.
  41. 5:41 And then most importantly also the reasoning.
  42. 5:44 So why an AI agent should do a certain task based on predefined policies and rules.
  43. 5:56 So, and behind this story is always our, let's say the foundation, if we would like to build agentic GraphRag applications.
  44. 6:06 We always have the AI agent, where the user sends first a query.
  45. 6:15 The agent looks for this knowledge, or for this topic, if it's available in its knowledge source.
  46. 6:22 If not, we,
  47. 6:24 jump to the graph database, we have a certain asset of tools like text to cipher that translates text, human text to our query language and then we traverse the graph on the right content and hopefully coming back to the user with more reliable and qualitative content.
  48. 6:48 So, ABK, you have been building a decision framework in the last couple of days.
  49. 6:54 Let's have a look on that.
  50. 6:55 Sure.

Open at this second