AI Video Analysis

Upload a video.
AI tells you everything.

Drop a video file or paste a YouTube URL — Kunya processes every frame, every word spoken, every visual detail. Get detailed analysis, answers to your questions, or structured data for automation. Use it in chat, through the API, or with autonomous agents.

product review
q3-product-demo.mp4
4:32

"Detailed notes on this demo — features, issues, and impressions."

Video AnalysisComplete

Features Demonstrated

The demo covers three core features: the dashboard redesign (0:00–1:45), search with filters and saved queries (1:46–3:10), and settings with role-based permissions (3:11–4:32). Each is shown with a live demo against production data.

Issues Spotted

At 1:24, a layout shift when the sidebar collapses — content jumps left ~20px. At 3:07, "Organization Admin" is truncated in the permissions table. At 2:38, a loading spinner takes ~3s — likely a slow API call the presenter skips past.

Presenter Delivery

Confident, clear narration. Good pacing between features. Could improve transitions — the jump from search to settings feels abrupt at 3:10.

Recommendation

Fix the layout shift and text truncation before sharing externally. Consider adding a brief recap slide at the end. Overall: ready for internal review, needs minor polish for customer-facing use.

swing analysis
golf-swing-slo-mo.mov
0:18

"Analyze my golf swing. What am I doing wrong and how can I improve?"

Swing AnalysisComplete

Grip & Address

Your grip looks solid — neutral with the V's pointing toward the right shoulder. At address (0:00–0:03), your stance is slightly narrow for a driver. Widening by ~2 inches would improve stability through the swing.

Backswing

Good shoulder turn reaching about 90°. However, at 0:05 your left arm bends noticeably at the top — this reduces power and consistency. Your hips also over-rotate past 45°, which leaks stored energy.

Downswing & Impact

The downswing initiates well from the lower body (0:08). At impact (0:11), your hands are slightly behind the ball — you're scooping rather than compressing. This causes the high launch angle and distance loss.

Key Drills

1. Towel drill for arm extension. 2. Hip bump drill to limit backswing rotation. 3. Impact bag work to train forward shaft lean. Focus on these for 2 weeks and re-record.

security audit
lobby-cam-04-19.mp4
1:42:00

"Review this lobby footage. Flag anything unusual — unauthorized access, tailgating, unattended items."

Security Review3 Flags

Overview

1 hour 42 minutes of lobby footage. 47 individuals entered the building. 44 badged in normally. 3 incidents flagged for review.

Flag 1 — Tailgating

High

At 0:23:14, an individual in a gray jacket follows an employee through the badge-controlled door without scanning. They walk directly to the elevator bank. No visitor badge visible.

Flag 2 — Unattended Bag

Medium

A black backpack is left on the bench near reception at 0:51:02. It remains unattended for 14 minutes before being retrieved by the same person who left it.

Flag 3 — After-Hours Entry

Medium

At 1:37:45, an employee badges in after normal hours (past 7 PM). They access the server room corridor at 1:38:20. Badge reads as "J. Martinez — Engineering." May be legitimate but worth cross-referencing with access logs.

Input Methods

Any video. Any source.

Multiple ways to get your video into Kunya for analysis.

File Upload

Drag and drop video files directly. MP4, MOV, WebM, AVI, and more — no conversion needed.

YouTube URLs

Paste any YouTube link. Kunya accesses the video directly — no downloading, no file size limits.

In-Chat Attachments

Attach videos directly in any AI chat conversation. Ask questions about the video as naturally as chatting.

REST API

Send video URLs or base64-encoded files programmatically. Build video analysis into your own products.

What AI Sees

Complete multimodal comprehension.

Not just frame extraction — true understanding of visual content, speech, text, motion, and temporal context working together.

Visual Understanding

Objects, scenes, actions, people, composition, spatial relationships, color, lighting — everything visible in every frame.

Speech & Dialogue

Transcribes and understands spoken words, tone, emphasis, speaker changes, and conversational context.

On-Screen Text

Reads titles, captions, code, UI labels, slides, charts, and any text that appears on screen across frames.

Motion & Action

Tracks movement patterns, body language, gestures, camera movements, transitions, and temporal sequences.

Audio Context

Music, sound effects, ambient noise, applause, silence — full audio-visual correlation for complete understanding.

Temporal Reasoning

Understands before/after relationships, cause-and-effect, narrative flow, pacing, and time-based patterns.

Applications

Real-world use cases.

Video analysis unlocks insights that text and images alone cannot provide.

Sports & Fitness Coaching

Analyze technique, form, and movement patterns. Get frame-level feedback on swings, strokes, jumps, and footwork.

Content Creation & Strategy

Study competitor videos, extract hooks, analyze pacing and editing patterns. Understand what makes content perform.

QA & Bug Detection

Find visual bugs, animation glitches, layout shifts, and rendering artifacts in UI recordings automatically.

Education & Training

Review lectures, extract key moments, assess delivery, create study guides, and generate quizzes from video.

Security & Compliance

Monitor footage for anomalies, audit recordings for compliance violations, detect unauthorized access patterns.

Product & UX Research

Analyze user session recordings, document competitor features, compare interface flows, extract UX patterns at scale.

Medical & Scientific

Analyze lab recordings, surgical technique reviews, microscopy footage, and experimental observations.

Legal & Documentation

Transcribe depositions, analyze evidence footage, document incident recordings, extract key moments for case files.

Media & Broadcasting

Auto-generate highlights, detect scene changes, create chapter markers, classify content, and build searchable archives.

How It Works

From video to insight in seconds.

01

Send your video

Upload a file, paste a YouTube URL, or attach a video in chat. The API accepts URLs and base64 files.

02

AI processes everything

Advanced multimodal models analyze every visual frame, audio track, spoken word, and on-screen text simultaneously.

03

Get actionable results

Receive natural language analysis, structured JSON for automation, or timestamped insights — however you need it.

Output Formats

Results your way. Your format.

Get video analysis results in the format that fits your workflow.

Structured JSON

Define your own output schema. Get typed, validated JSON ready for pipelines, webhooks, and databases.

Natural Language

Ask questions and get conversational answers. Follow up, drill deeper, compare sections — like talking to a colleague.

Timestamped Insights

Key moments, scene changes, and events with precise timestamps. Navigate directly to what matters.

Full Summaries

Complete video breakdowns with sections, highlights, key takeaways, and action items.

Video Formats

Works with everything.

MP4
MOV
WebM
AVI
MKV
YouTube URLs

Up to 2 hours. Any resolution. Any frame rate.

Available Everywhere

Three ways to use video analysis.

In Chat

Attach a video to any conversation. Ask follow-up questions, compare videos, request different analysis formats — as natural as chatting.

Via API

POST a video URL or file to the /v1/videos/analyze endpoint. Get structured JSON back. Build video intelligence into your apps.

Through Agents

Kunya agents can autonomously analyze videos as part of larger workflows — content audits, QA pipelines, research tasks, and more.

Developer API

Build video intelligence into anything.

One API call. Any video. Structured results. The same models powering the chat, available programmatically.

analyze-video.sh

curl -X POST https://api.kunya.ai/v1/videos/analyze \

-H "Authorization: Bearer kunya_..." \

-H "Content-Type: application/json" \

-d '{

"video_url": "https://youtube.com/watch?v=...",

"prompt": "Give me detailed notes on this demo",

"response_format": "text"

}'

OpenAI-compatible endpoint
YouTube URL or file upload
Custom output schemas
Streaming responses
Webhook callbacks
Usage metered per call

Upload a video. See what AI sees.

No credit card required. Analyze your first video in seconds.