> ## Documentation Index
> Fetch the complete documentation index at: https://docs.trugen.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Hawkeye-1

> TruGen's Vision Action Recognition Model — the perception layer that gives AI agents the ability to see, hear, and understand humans in context.

**Hawkeye-1 is TruGen's Vision Action Recognition Model** — a real-time perception system that goes far beyond conventional emotion detection. Instead of reducing human state to a handful of categories, Hawkeye-1 reads body language, vocal tone, subtle facial expressions, environmental cues, and conversational timing simultaneously to build a layered understanding of the person on the other side of the conversation.

<div style={{ overflow: "hidden", borderRadius: "12px", lineHeight: 0, marginLeft: "-1.5rem", marginRight: "-1.5rem", marginTop: "1rem", marginBottom: "1rem" }}>
  <img src="https://mintcdn.com/trugen/9qi__syJVXNUbiUr/images/hawkeye-1.png?fit=max&auto=format&n=9qi__syJVXNUbiUr&q=85&s=7cb907cfbee52d781ef398e397036f18" alt="Hawkeye-1 Vision Action Recognition Model" style={{ width: "100%", display: "block" }} width="672" height="325" data-path="images/hawkeye-1.png" />
</div>

## The three pillars of Hawkeye-1

### Multimodal perception

Hawkeye-1 sees and listens in parallel. It combines what it observes on camera with what it hears in speech to form a unified, multimodal understanding of the user. This is what makes it a true perception system, not just a computer-vision model.

### Environmental awareness

Context doesn't stop at the face. Hawkeye-1 continuously analyses the video feed, tracking environmental changes, detecting key gestures and behaviours, and triggering relevant actions as needed. Whether a user holds up a product for a visual question, steps away from the screen, or changes their surroundings, Hawkeye notices — and the agent adapts.

### Emotional intelligence

Hawkeye-1 reads body language, vocal tone, and subtle facial expressions to render emotional understanding the way humans do — with real-time awareness of context and conversational flow. The result is an agent that doesn't just react to words, but responds to meaning.

## Developer-friendly by design

Hawkeye-1 is built for integration. A single API flag enables vision perception, and natural-language prompts let you configure what to track — objects, gestures, on-screen activity, or all of the above.

* **Enable Hawkeye with a single flag** — flip one parameter and your agent gains real-time scene analysis, effortlessly and precisely.
* **Answer visual questions** — the model sees and understands the user's environment, responding to visual queries about objects, scenes, and actions just like a human would.
* **Custom action triggers** — automate responses and fire tool calls at the right moment based on what Hawkeye perceives (timing, tone, gesture, or environment).

## What you can build with Hawkeye-1

<CardGroup cols={2}>
  <Card title="Healthcare" icon="stethoscope">
    Understand patient mood and engagement in real time. Trigger tool calls at the right moment based on timing, tone, and conversational flow.
  </Card>

  <Card title="Education & training" icon="graduation-cap">
    Avatars that gauge learner engagement, adjust pacing, and provide personalised guidance — making training more effective at scale.
  </Card>

  <Card title="Customer service" icon="headset">
    Handle FAQs and guide users through complex flows while reading frustration or confusion before it escalates.
  </Card>

  <Card title="Sales" icon="chart-line">
    Qualify leads and showcase products with avatars that read buyer signals — hesitation, interest, objection — and adapt in real time.
  </Card>

  <Card title="Recruiting" icon="clipboard-question">
    Screen candidates with adaptive interviews that evaluate answers, communication style, confidence, and engagement.
  </Card>
</CardGroup>

## Better together: Hawkeye-1 + Huma

Hawkeye-1 is powerful on its own, but reaches its full potential when paired with the [Huma family](/docs/models/overview) of avatar renderers. Together they deliver:

* **Visual realism + situational awareness** — Huma renders photorealistic faces; Hawkeye-1 adds real-time perception of who's watching.
* **Emotionally aware + contextually responsive** — Huma expresses emotion through micro-expressions; Hawkeye-1 tells it exactly when and how.
* **Real-time emotional sync** — expression, rhythm, and context synchronise into a single seamless conversational experience.

The combination turns static chatbots into living, breathing video agents — capable of seeing, understanding, and responding with the nuance of a real human conversation.

## Key characteristics

| Attribute  | Detail                                                             |
| ---------- | ------------------------------------------------------------------ |
| Modality   | Continuous vision perception (camera + audio, in parallel)         |
| Input      | Live user webcam and screen share                                  |
| Output     | Structured scene, gesture, and emotional context passed to the LLM |
| Detection  | Emotion, gaze, head pose, face count, gestures, environment        |
| Activation | Single API flag on the agent configuration                         |

## Next steps

<CardGroup cols={2}>
  <Card title="Vision Understanding" icon="eye" href="/docs/agents/vision/overview">
    How to enable Hawkeye-1 on your agent and configure what to track.
  </Card>

  <Card title="Models overview" icon="brain" href="/docs/models/overview">
    See how Hawkeye-1 fits alongside the Huma family of renderers.
  </Card>
</CardGroup>