{
  "docId": "019dd923-5e88-73ef-bd5d-0d0f98caffe1",
  "docSlug": "5df6bd1b0447b5f6",
  "documentTitle": "2025 Air Street Capital The State of AI Report 2025",
  "authorId": "AirStreetCapital",
  "authorName": "Air Street Capital",
  "documentKindSlug": "consulting-deck",
  "documentKindLabel": "Consulting deck",
  "sourceTypeSlug": "vc_research",
  "sourceTypeLabel": "VC research",
  "presentationDate": null,
  "orientation": "landscape",
  "aspectRatio": 1.777,
  "pageNumber": 83,
  "pageCount": 313,
  "prevPage": 82,
  "nextPage": 84,
  "slideType": "industry_trends",
  "function": "present_solution",
  "density": "overcrowded",
  "nDataPoints": 4,
  "notes": "The slide presents a technical argument for using smaller models (1-9B parameters) in agentic workflows to reduce costs and latency, while maintaining an 'escape hatch' to larger models.",
  "elementsJson": [
    "headline_text",
    "action_title",
    "bullet_list",
    "process_diagram"
  ],
  "metadataConfidence": 0.95,
  "imagePath": null,
  "slideHref": "/slides/019dd923-5e88-73ef-bd5d-0d0f98caffe1/83",
  "deckHref": "/decks/019dd923-5e88-73ef-bd5d-0d0f98caffe1",
  "deckJsonHref": "/decks/019dd923-5e88-73ef-bd5d-0d0f98caffe1.json",
  "deckAnchorHref": "/decks/019dd923-5e88-73ef-bd5d-0d0f98caffe1#slide-83",
  "components": [
    {
      "bbox": null,
      "kind": "callout",
      "text": "They recommend SLM-first, heterogeneous agents that invoke large models only when needed.",
      "attrs": null,
      "subkind": null,
      "toolName": "Visual emphasis",
      "toolSlug": "visual-emphasis",
      "confidence": null,
      "componentId": "019dd952-41ac-7180-a822-b79e4e25a4f6",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.6,
        "w": 0.45,
        "x": 0.53,
        "y": 0.35
      },
      "kind": "image",
      "text": "Diagrams showing agent control flows with LMs and tools.",
      "attrs": null,
      "subkind": "infographic",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "34660419-5fc9-48ee-855c-5a7dcc7934fe",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.08,
        "w": 0.48,
        "x": 0.02,
        "y": 0.84
      },
      "kind": "list",
      "text": "But SLMs still struggle with long context, novel reasoning, or messy conversation. Keep an escape hatch to a large model and evaluate regularly.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "150bdca3-955b-4660-83dd-8cabc3d063f5",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.12,
        "w": 0.48,
        "x": 0.02,
        "y": 0.55
      },
      "kind": "list",
      "text": "A ~7B model is typically 10-30x cheaper to run and responds faster. You can fine-tune it overnight with LoRA/QLoRA and even run it on a single GPU or device.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "69deaa52-8a2b-43f6-a055-6c9d5d0b0a8a",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.15,
        "w": 0.48,
        "x": 0.02,
        "y": 0.38
      },
      "kind": "list",
      "text": "Agents mostly fill forms, call APIs, follow schemas, and write short code. New small models (1–9B) do these jobs well: Phi-3-7B and DeepSeek-R1-Distill-7B handle instructions and tools competitively.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "9fba291c-d658-4e33-b744-8a178bca1753",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.15,
        "w": 0.48,
        "x": 0.02,
        "y": 0.68
      },
      "kind": "list",
      "text": "One can use a “small-first, escalate if needed” design: route routine calls to an SLM and escalate only the hard, open-ended ones to a big LLM. In practice, this can shift 40-70% of calls to small models with no quality loss.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "f2ffdd87-9132-4eb5-a547-d1eee4231626",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": null,
      "kind": "metric",
      "text": "Cost reduction: 10-30x",
      "attrs": null,
      "subkind": "primary",
      "toolName": "Quantification",
      "toolSlug": "quantification",
      "confidence": null,
      "componentId": "019dd952-41ac-7180-a822-ba30b1a9a75b",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.15,
        "w": 0.48,
        "x": 0.02,
        "y": 0.22
      },
      "kind": "paragraph",
      "text": "Researchers from NVIDIA and Georgia Tech argue that most agent workflows are narrow, repetitive, and format-bound, so small language models (SLMs) are often sufficient, more operationally suitable, and much cheaper. They recommend SLM-first, heterogeneous agents that invoke large models only when needed.",
      "attrs": null,
      "subkind": "paragraph",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "09203834-0dd8-4dc6-94b6-e1d927154ae5",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.05,
        "w": 0.5,
        "x": 0.02,
        "y": 0.14
      },
      "kind": "title",
      "text": "Could small language models be the future of agentic AI?",
      "attrs": null,
      "subkind": "headline",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "14a0e7dc-b5c9-47c8-87b4-829c4e303604",
      "frameworkName": null,
      "frameworkSlug": null
    }
  ],
  "metrics": [],
  "tools": [],
  "frameworks": [],
  "arcBeats": [
    {
      "to": 189,
      "from": 12,
      "beatId": "019dd95a-0682-776c-8e35-9f380673831f",
      "arcName": "The Triple Take",
      "arcSlug": "triple-take",
      "beatName": "The Facts (What)",
      "beatSlug": "triple-take-the-facts-what",
      "evidence": "Sections 1-2 lay out research findings and industry data with charts and case studies.",
      "position": 1,
      "confidence": 70,
      "parentBeatName": "Setup",
      "parentBeatSlug": "setup"
    },
    {
      "to": 89,
      "from": 12,
      "beatId": "019dd95a-0682-776c-8e35-ac68557d958b",
      "arcName": "The Onion",
      "arcSlug": "onion",
      "beatName": "First Layer",
      "beatSlug": "onion-first-layer",
      "evidence": "Research section unpacks technical layer beneath the headlines.",
      "position": 2,
      "confidence": 45,
      "parentBeatName": "Development",
      "parentBeatSlug": "development"
    }
  ],
  "loops": [],
  "imagePathAlt": null,
  "thumbSrc": null,
  "thumbSrcAlt": null,
  "locked": true
}