postreef
All schemas

Conference Talk

vidextract.predefined.talk.v1

A conference keynote, tech talk, or recorded presentation extracted from a video. Capture the thesis, structure, and takeaways the speaker delivers. Record only what is stated; omit anything not mentioned and do not infer.

Fields

titlerequired
string

Title of the talk, as stated or shown.

speaker
string

Name of the presenter. Omit if not stated.

speakerTitle
string

Speaker's role, company, or affiliation, e.g. 'Staff Engineer at Acme'. Omit if not stated.

event
string

Name of the conference or event, e.g. 'NDC London 2026'. Omit if not stated.

thesisrequired
string

The central argument or main message of the talk, in one to three sentences.

topics
string[]

Subjects, technologies, or themes covered, lowercased, e.g. 'large language models', 'distributed systems'.

audience
string

Intended audience as stated, e.g. 'software engineers', 'engineering leaders'. Omit if not stated.

keyPointsrequired
object[]

Main points or sections of the talk, in the order presented.

keyTakeaways
string[]

Actionable conclusions or lessons the audience should leave with. Short sentences.

recommendations
string[]

Specific advice, best practices, or calls to action the speaker gives. Short phrases.

toolsMentioned
string[]

Specific tools, products, frameworks, or libraries named, e.g. 'GitHub Copilot', 'Kubernetes'.

quotes
string[]

Memorable or quotable lines from the speaker, verbatim where possible.

predictions
string[]

Forward-looking claims or predictions the speaker makes about the future. Short sentences.

JSON Schema

{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "vidextract.predefined.talk.v1",
  "title": "Conference Talk",
  "description": "A conference keynote, tech talk, or recorded presentation extracted from a video. Capture the thesis, structure, and takeaways the speaker delivers. Record only what is stated; omit anything not mentioned and do not infer.",
  "type": "object",
  "required": [
    "title",
    "thesis",
    "keyPoints"
  ],
  "additionalProperties": false,
  "properties": {
    "title": {
      "type": "string",
      "description": "Title of the talk, as stated or shown."
    },
    "speaker": {
      "type": "string",
      "description": "Name of the presenter. Omit if not stated."
    },
    "speakerTitle": {
      "type": "string",
      "description": "Speaker's role, company, or affiliation, e.g. 'Staff Engineer at Acme'. Omit if not stated."
    },
    "event": {
      "type": "string",
      "description": "Name of the conference or event, e.g. 'NDC London 2026'. Omit if not stated."
    },
    "thesis": {
      "type": "string",
      "description": "The central argument or main message of the talk, in one to three sentences."
    },
    "topics": {
      "type": "array",
      "description": "Subjects, technologies, or themes covered, lowercased, e.g. 'large language models', 'distributed systems'.",
      "items": {
        "type": "string"
      }
    },
    "audience": {
      "type": "string",
      "description": "Intended audience as stated, e.g. 'software engineers', 'engineering leaders'. Omit if not stated."
    },
    "keyPoints": {
      "type": "array",
      "description": "Main points or sections of the talk, in the order presented.",
      "minItems": 1,
      "items": {
        "type": "object",
        "required": [
          "point"
        ],
        "additionalProperties": false,
        "properties": {
          "point": {
            "type": "string",
            "description": "The key point or claim, e.g. 'AI coding tools shift the bottleneck from writing to reviewing code.'"
          },
          "detail": {
            "type": "string",
            "description": "Supporting explanation, example, or evidence the speaker gives. Omit if self-contained."
          },
          "videoTimestamp": {
            "type": "string",
            "description": "Start timestamp in source video, 'mm:ss' or 'h:mm:ss'. Omit if unknown.",
            "pattern": "^(\\d{1,2}:)?\\d{1,2}:\\d{2}$"
          }
        }
      }
    },
    "keyTakeaways": {
      "type": "array",
      "description": "Actionable conclusions or lessons the audience should leave with. Short sentences.",
      "items": {
        "type": "string"
      }
    },
    "recommendations": {
      "type": "array",
      "description": "Specific advice, best practices, or calls to action the speaker gives. Short phrases.",
      "items": {
        "type": "string"
      }
    },
    "toolsMentioned": {
      "type": "array",
      "description": "Specific tools, products, frameworks, or libraries named, e.g. 'GitHub Copilot', 'Kubernetes'.",
      "items": {
        "type": "string"
      }
    },
    "quotes": {
      "type": "array",
      "description": "Memorable or quotable lines from the speaker, verbatim where possible.",
      "items": {
        "type": "string"
      }
    },
    "predictions": {
      "type": "array",
      "description": "Forward-looking claims or predictions the speaker makes about the future. Short sentences.",
      "items": {
        "type": "string"
      }
    }
  }
}

Use it via the API

Pass schemaId: "vidextract.predefined.talk.v1" when you submit an extraction. The call returns an id immediately. Poll GET /v1/extractions/{id}/result (or register a webhook) for the structured result.

curl -s -X POST https://postreef.com/v1/extractions \
  -H "x-api-key: $VIDEXTRACT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"url": "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
       "inputs": ["transcript", "comments"],
       "schemaId": "vidextract.predefined.talk.v1"}'

Full request/response shape in the API reference; copy-paste poll loops in the API quickstart.

Pass schemaId: "vidextract.predefined.talk.v1" in your request to extract with this schema.