---
schema: 1
type: video
title: 'Enterprise Agents Have a Structure Problem - Ishita Daga, Tesla'
description: >-
  Most enterprise agents fail for the same reason: the model can generate SQL,
  call tools, and follow workflows, but it has no understanding of how the
  business actually defines its data. Most teams try to fix this with longer
  prompts, more RAG, or a bigger model. The real fix is building semantic
  retrieval infrastructure — a machine-readable metadata layer that lets agents
  reason over business concepts instead of guessing at raw schemas. In this
  talk, I’ll walk through how this metadata powers business-context-aware agents
  through semantic retrieval, metadata graphs, and domain-specific sub-agents.
  Speakers: - Ishita Daga (Tesla): Ishita Daga is a Senior Machine Learning
  Engineer at Tesla, building enterprise AI systems that combine semantic
  retrieval, metadata infrastructure, and business-aware reasoning to make
  analytics agents reliable in production. Previously, she was an AI Scientist
  at Covera Health, where she built multimodal and weakly supervised ML systems
  for radiology intelligence. Her interests include enterprise agent
  architectures, semantic layers, and operational AI systems.
canonical: >-
  https://liquidgold.cc/channel/aidotengineer/explore/video/enterprise-agents-have-a-structure-problem-ishita-d-0d69ed0e1b87
markdown: >-
  https://liquidgold.cc/channel/aidotengineer/video/enterprise-agents-have-a-structure-problem-ishita-d-0d69ed0e1b87.md
llms: null
parent: 'https://liquidgold.cc/channel/aidotengineer.md'
channel: aidotengineer
updated: '2026-10-24T07:13:22.723Z'
language: null
children: []
videoId: B8l81jhvHbI
durationSeconds: 728
publishedAt: '2026-07-20T06:25:12.000Z'
searchChunks:
  - ordinal: 0
    text: null
    startSeconds: 13
    endSeconds: 121
    watch: 'https://liquidgold.cc/watch/B8l81jhvHbI?t=13'
    heldReason: rights
  - ordinal: 1
    text: null
    startSeconds: 13
    endSeconds: 121
    watch: 'https://liquidgold.cc/watch/B8l81jhvHbI?t=13'
    heldReason: rights
  - ordinal: 2
    text: null
    startSeconds: 13
    endSeconds: 121
    watch: 'https://liquidgold.cc/watch/B8l81jhvHbI?t=13'
    heldReason: rights
  - ordinal: 3
    text: null
    startSeconds: 468
    endSeconds: 581
    watch: 'https://liquidgold.cc/watch/B8l81jhvHbI?t=468'
    heldReason: rights
  - ordinal: 4
    text: null
    startSeconds: 133
    endSeconds: 248
    watch: 'https://liquidgold.cc/watch/B8l81jhvHbI?t=133'
    heldReason: rights
  - ordinal: 5
    text: null
    startSeconds: 248
    endSeconds: 260
    watch: 'https://liquidgold.cc/watch/B8l81jhvHbI?t=248'
    heldReason: rights
  - ordinal: 6
    text: null
    startSeconds: 260
    endSeconds: 343
    watch: 'https://liquidgold.cc/watch/B8l81jhvHbI?t=260'
    heldReason: rights
  - ordinal: 7
    text: null
    startSeconds: 354
    endSeconds: 456
    watch: 'https://liquidgold.cc/watch/B8l81jhvHbI?t=354'
    heldReason: rights
  - ordinal: 8
    text: null
    startSeconds: 354
    endSeconds: 456
    watch: 'https://liquidgold.cc/watch/B8l81jhvHbI?t=354'
    heldReason: rights
  - ordinal: 9
    text: null
    startSeconds: 468
    endSeconds: 581
    watch: 'https://liquidgold.cc/watch/B8l81jhvHbI?t=468'
    heldReason: rights
  - ordinal: 10
    text: null
    startSeconds: 591
    endSeconds: 728
    watch: 'https://liquidgold.cc/watch/B8l81jhvHbI?t=591'
    heldReason: rights
  - ordinal: 11
    text: null
    startSeconds: 591
    endSeconds: 728
    watch: 'https://liquidgold.cc/watch/B8l81jhvHbI?t=591'
    heldReason: rights
---


# Enterprise Agents Have a Structure Problem - Ishita Daga, Tesla

Most enterprise agents fail for the same reason: the model can generate SQL, call tools, and follow workflows, but it has no understanding of how the business actually defines its data. Most teams try to fix this with longer prompts, more RAG, or a bigger model. The real fix is building semantic retrieval infrastructure — a machine-readable metadata layer that lets agents reason over business concepts instead of guessing at raw schemas. In this talk, I’ll walk through how this metadata powers business-context-aware agents through semantic retrieval, metadata graphs, and domain-specific sub-agents. Speakers: - Ishita Daga (Tesla): Ishita Daga is a Senior Machine Learning Engineer at Tesla, building enterprise AI systems that combine semantic retrieval, metadata infrastructure, and business-aware reasoning to make analytics agents reliable in production. Previously, she was an AI Scientist at Covera Health, where she built multimodal and weakly supervised ML systems for radiology intelligence. Her interests include enterprise agent architectures, semantic layers, and operational AI systems.

Canonical: https://liquidgold.cc/channel/aidotengineer/explore/video/enterprise-agents-have-a-structure-problem-ishita-d-0d69ed0e1b87 · Channel: [AI Engineer](https://liquidgold.cc/channel/aidotengineer.md)

> Search chunk text is held until public-source provenance is verified.

## Document chunks

---

#### Chunk 0: Document excerpt (0:01 to 2:01)

> Held: transcript rights do not permit this chunk's text.

---

#### Chunk 1: Document excerpt (2:01 to 4:08)

> Held: transcript rights do not permit this chunk's text.

---

#### Chunk 2: Document excerpt (4:08 to 5:43)

> Held: transcript rights do not permit this chunk's text.

---

#### Chunk 3: Document excerpt (5:43 to 7:36)

> Held: transcript rights do not permit this chunk's text.

---

#### Chunk 4: Document excerpt (7:36 to 9:41)

> Held: transcript rights do not permit this chunk's text.

---

#### Chunk 5: Document excerpt (9:41 to 12:08)

> Held: transcript rights do not permit this chunk's text.