Skip to main content
Memories.ai ships three independent products on top of one shared video understanding stack — plus a CLI that drives all of them from the terminal. Pick the product that matches what you’re building.

​
Get Started

Create Your API Key

Generate one sk-mavi-... key that works across Visual Intelligence, Visual Search, and Visual Agents. Under 2 minutes.

Upload Your First Video

The Visual Search indexing pipeline at a glance — pick the right upload method and learn how to wait for the parse to finish.

​
Products

Visual Intelligence

Stateless inference APIs. Direct REST calls for transcription, captioning, model inference, embeddings, live-stream moderation, and Human ReID. No persistence — you bring the data, we return the result.

Visual Search

Indexed video + image library. Upload videos and images once, the platform auto-indexes them, and you query by natural language, image similarity, or transcript phrase. State lives on the server.

Visual Agents

Pre-built agents + workflow APIs. Open-source video-searching and video-editing agents, plus managed services for queries, clip/edit/split, and screenplay extraction. Fork the agents, or call the managed endpoints directly.

​
Which One Do I Use?

If you want to…Use
Run AI analysis on a one-off video file you haveVisual Intelligence — Video Model APIs or Video Task APIs
Transcribe an audio or video file to textVisual Intelligence — Audio File Transcription
Pull a transcript / caption from a YouTube / TikTok / Instagram / X linkVisual Intelligence — Social Media Scraping
Moderate a live RTMP stream in real timeVisual Intelligence — Live Video Content Moderation
Transcribe a live audio broadcast in real timeVisual Intelligence — Live Audio Transcription
Identify specific named people in a videoVisual Intelligence — Human ReID & Caption
Build a searchable video library you can query laterVisual Search — upload once, auto-indexed, query later
Find moments across all your uploaded videos by natural languageVisual Search — Search by Text
Build a video discovery / editing botVisual Agents — Video Searching Agent or Video Editing Agent
Extract storyboard / screenplay data from short drama episodesVisual Agents — Screenplay Extraction
Drive the platform from the terminalMemories CLI — see Tools below

​
What’s Inside Each Product

​
Visual Intelligence

Stateless REST APIs on https://mavi-backend.memories.ai/serve/api/v2 (plus two specialty hosts — see Base URLs).
GroupWhat it does
Asset ManagementUpload / download / delete video and image assets used by other VI APIs
Social Media ScrapingMetadata, transcripts, captions, comments from YouTube, Instagram, TikTok, Twitter/X
Audio File TranscriptionWhisper, ElevenLabs, AssemblyAI providers + speaker diarization / recognition
Live Audio TranscriptionReal-time STT on live audio — server-pull (callback) or WebSocket (client-push)
Video Model APIsDirect VLM calls with your own prompt — Gemini, Nova, Qwen
Video Task APIsPre-packaged tasks on top of VLMs — Video Frame Description, Video Summary
Live Video Content ModerationNSFW / violence / logo detection on RTMP/RTSP streams
Live Video UnderstandingCustom AI prompt continuously applied to a live RTMP stream
Image Model APIsDirect ILM calls — Gemini, GPT, Nova, Qwen
EmbeddingsImage / video / text embeddings for semantic search and retrieval
Human ReID & CaptionIdentity-aware vision — caption a video with named people. Requires a dedicated key
Example — transcribe an audio URL: 
curl -X POST "https://mavi-backend.memories.ai/serve/api/v2/transcriptions/speech-to-text" \
  -H "Authorization: sk-mavi-..." \
  -H "Content-Type: application/json" \
  -d '{
    "provider": "elevenlabs",
    "url": "https://example.com/recording.mp3",
    "language_code": "en"
  }'
Indexed video + image library on https://api.memories.ai/serve/api/v1. Upload once, query forever.
GroupWhat it does
Index UploadUpload videos (from file, URL, or social-media creator handle) and images for indexing
SearchSemantic and keyword search across your private library and the public video library — by text, image, or transcript phrase
Library ManagementList / get metadata / download / delete videos in your library
Example — upload a video for indexing: 
import requests

with open("meeting.mp4", "rb") as f:
    response = requests.post(
        "https://api.memories.ai/serve/api/v1/upload",
        headers={"Authorization": "sk-mavi-..."},
        files={"file": ("meeting.mp4", f, "video/mp4")},
        data={"unique_id": "my-project", "callback": "https://your.app/callback"},
    )
print(response.json())  # returns videoNo for later search

​
Visual Agents

Reference implementations and managed workflow APIs on https://mavi-backend.memories.ai/serve/api/v2.
Agent / ServiceOpen-source repoManaged API
Video Searching Agentvideo-searching-agentPOST /queries/stream
Video Editing Agent (VEA)vea-open-sourcePOST /video/edit, /video/clip, /video/split
Screenplay Extraction— (managed only)POST /screenplay/tasks (async)
Example — stream a Video Searching Agent query: 
curl -N -X POST "https://mavi-backend.memories.ai/serve/api/v2/queries/stream" \
  -H "Authorization: sk-mavi-..." \
  -H "Content-Type: application/json" \
  -d '{
    "query": "Trending AI coding tool videos on TikTok this week",
    "platforms": ["tiktok"],
    "max_results": 5,
    "time_frame": "past_week"
  }'

​
Tools

​
Memories CLI

Command-line tool for the entire ecosystem — uploads, searches, agent calls, asset management from your terminal or shell scripts. Same sk-mavi-... key as the rest of the platform.

Memories CLI on GitHub

Install instructions, command reference, and source.

Memories.ai Console

Manage API keys, view usage and credits, configure webhooks.

​
Billing

All three products share a unified billing system. Pricing varies per endpoint — see the pricing note at the top of each endpoint page, or the Memories.ai Console for your current usage and credits.

​
Compliance

Memories.ai meets the highest standards of data security and privacy with HIPAA, SOC 2 Type 2, and GDPR compliance. For more information, see the Trust Center.