Introducing the world’s first Large Visual Memory Model - AI that sees and remembers video like ChatGPT does with text. Upload to Memories.ai and search or ask questions anytime.
I’m Shawn, founder of Memories.ai, former researcher at Meta, and PhD in Computer Science from the University of Cambridge.
🚀 Today we’re launching Memories.ai – the world’s first Large Visual Memory Model (LVMM). Our goal? Give AI human-like visual memory.
🧠 Why Visual Memory?
AI today is mostly chat-based. But humans aren’t just chat-based – we think, remember, and interact with the world visually. Visual memory is core to how we understand people, places, and context. That’s what we’re bringing to AI.
With Memories.ai, AI can see, understand, and remember your video content – just like a person would. Once your videos are uploaded and indexed, you can interact with them naturally, ask questions, and retrieve any moment instantly — no need to reprocess anything.
Great idea, congrats on the launch! As some others have said, would love to use this for podcasts / YouTube content. I saw that's in the works, excited to try that out
TECHMATEHUB is your friendly spot for all things tech. From tips and tutorials to the latest gadget news, it’s built to help you stay ahead in the digital world.https://techmatehub.com/ Whether you’re a beginner or a tech lover, there’s something useful for everyone. Simple, smart, and always up to date!
FASTSYSTEMPATCH is a quick and easy solution for keeping your system running smoothly. It helps fix bugs, improve performance, and boost security with simple updates.https://fastsystempatch.com/ Designed for all users, it saves time and cuts the hassle of complex tech fixes. Just patch and go—fast and stress-free.
Ask ChatGPT
DAILYMASHUPHUB is your daily dose of trending news, entertainment, and viral content. It brings together stories from around the web in one fun, easy-to-browse place.https://dailymashuphub.com/ Whether you’re into pop culture, tech, or lifestyle, there’s always something fresh to check out. It’s a great spot to stay updated and entertained on the go.
ZAPTECHTOOLS is a go-to platform for handy tech tools and resources. Whether you're working on content, coding, or online tasks, it offers simple, easy-to-use solutions.https://zaptechtools.com/ The site is designed for speed and convenience, making your work smoother and faster. It's a great choice for anyone who wants reliable tools without the hassle.
With AI-generated content on the rise, having reliable tools to spot it is more important than ever.https://zerogpttool.com/ Whether you're a student, marketer, or business pro, making sure your content is original matters. ZeroGPTTool.com is one well-known option, but is it really the best? In this article, we’ll check out 22 other tools to see if ZeroGPTTool.com lives up to the hype—or if there are better choices out there.
Ventsbusiness has built a strong name in online publishing by making it easy for new businesses and creators to share their stories.https://ventsbusiness.co.uk/ With flexible options like news releases and feature articles, the platform helps startups and PR teams get noticed without the hurdles of traditional media. This open and user-friendly approach supports fresh voices and gives readers a mix of real stories and smart branding.
Congrats on the launch,
but I didn't get it, ain't you doing embedding in the backend from the VLM?
you are basically transcribing each frame of video in a multi-threaded way and after you get text you are indexing those with the PTS of video.
when user asks the question, it goes in the vector database and fetches the relevant frames based on the query with summary and timestamp.
A human on the other hand, can do lot more than that, humans can process visual input in real-time and then provide insights.
For Eg: I saw a panda fighting and training martial arts.
-- At one sight thats only I saw, but how carefully I have to look, depends. If somebody asks me "how many push-ups in total that panda did?". Then I need to look at the video again and count. But if you have processed it once, it impossible to get the detailed insights.
Best of Luck, btw its a great initiative. Might be I can have a wearable like a necklace or chain that has 24hrs batter life and records audio and video. Integrated with your platform can do much better.
Its like I can revisit my days from past.
Good Work!!!
This launch is fascinating! Memories.ai's ability to interact with video content just like a human is groundbreaking. I’m curious, how does the AI handle different languages in video content? Does it require subtitles for accurate understanding? 🤔
You creating these for to remove the chat bot system between human and ai , but before creating the video we have to chat with AI right ??
Congrats on launching Memories.ai! The ability for AI to remember and search videos like text is a game-changer!
this is a huge breakthrough and changes what can be achieved with ai apps. i can't wait to work with it
Congrats on the launch!
This is pretty cool. It could be great for curating educational video content or helping content creators and YouTubers surface specific scenes from their older videos.
I wouldn’t be surprised if one of the big foundational model players tries to acquire this. A memory layer for video understanding is kind of a layer they want to embed into their existing models.
Unlimited context is impressive! But:
1. Where is video processed? (On-device vs cloud)
2. How are temporal relationships stored? (e.g., 'Before X happened, Y occurred')
3. Can it distinguish identical objects across time? (e.g., 'Show all instances of this coffee mug')
Congrats @imshawnshen on the launch. Will check it out.
