Tavus

Tavus

Software Development

San Francisco, CA 8,112 followers

Build digital twin video experiences with easy-to-use APIs

About us

Tavus is a generative AI video research company that enables developers to build digital twin video experiences through easy-to-use APIs.

Website
http://www.tavus.io
Industry
Software Development
Company size
11-50 employees
Headquarters
San Francisco, CA
Type
Privately Held
Founded
2020

Locations

Employees at Tavus

Updates

  • Tavus reposted this

    View organization page for Cartesia, graphic

    3,071 followers

    We had a fantastic time hosting our Conversational AI Leaders Dinner for SF Tech Week last night. It’s always energizing to spend time with our community and hear how they’re pushing the boundaries of voice agents across such a wide range of industries. Thanks to this group for sharing their insights - go check out what they’re building! Tim Shi, CTO of Cresta - enterprise-grade AI for contact centers David Zhao, CTO of LiveKit - open source infrastructure for real time AI Sophia Xing, Former Head of Product Inworld AI - AI engine for gaming Quinn Favret, COO of Tavus - apis for real-time digital twins Jeffery L., Co-CEO of Assort Health - AI call centers for healthcare Peggy Wang, CTO of ego (YC W24)- AI-native 3D simulation engines Samir Sen, AI Engineer at Crescendo - AI powered customer service Derek Pankaew, CEO and Founder at Listening - text to audio for academic papers

    • No alternative text description for this image
    • No alternative text description for this image
  • View organization page for Tavus, graphic

    8,112 followers

    Happy #SFTechWeek! We've got 3 events happening this week 1️⃣ 🔥 Monday: Sneak peek a new model we're building. Sushobhan Ghosh will be sharing at GitHub: Register https://lu.ma/0mh3j8m1 2️⃣ 🍺 Tuesday: Join us, Zoom and Recall.ai for a casual Happy Hour at our office, meet others interested in AI Video: Register https://lu.ma/j7lqbug0 3️⃣ 🍽️ Wednesday: Private Dinner on AI Video (reach out to join waitlist)

    Awesome AI Dev Tools - October · Luma

    Awesome AI Dev Tools - October · Luma

    lu.ma

  • View organization page for Tavus, graphic

    8,112 followers

    One of our ML engineers saw this awkward lost luggage kiosk at the airport in Italy 😆 Imagine if you could actually talk to AI video agents like this at the airport. Ask them any question - about your flight, lost luggage, where to get food. Who's building interactive kiosks? With Tavus APIs you could build this today 😎

  • View organization page for Tavus, graphic

    8,112 followers

    Another cool and creative experiment from Reid Hoffman with his digital twin – this time from from the past! There is a small catch – there's clever editing used to make it look live. If you want to build an app or prototype with live conversational video, you can already do this with Tavus APIs. We're the fastest, with <1 second of latency between responses, making it feel pretty real. It's plug and play...get started in 5 mins for free: https://lnkd.in/gzWNf4c3

    View profile for Reid Hoffman, graphic
    Reid Hoffman Reid Hoffman is an Influencer

    Co-Founder, LinkedIn & Inflection AI. Investor at Greylock.

    Twenty years ago, I was about to pitch LinkedIn’s Series B to Greylock and other investors. I remember going through slides at Starbucks, equipped with a PowerPoint deck, a game plan for growth, and what I thought was a solid pitch. But there were, of course, challenges and uncertainties to address. More than just building a company, our team was trying to create a category. Some were calling us "Friendster for business," but the concept of professional networking was more or less untested. Our pitch deck had a vision I believed in deeply—enough to stake my career on it—but that didn’t make it easy. Knowing what I know now, what I would tell myself back in 2004, during that pivotal moment for me and for LinkedIn? What advice would I give younger Reid? What reassurances or warnings would I deliver? I’ve created a new Reid AI—20 years younger—to pitch the real, present-day me on the 2004 LinkedIn plan. To clarify: What you are watching is an attempt to use the latest AI tools to replicate some of my younger self’s curiosity and revisit an exciting moment for the company I co-founded—which is also an interesting test of how this technology can be used more generally. When we created “Reid AI”—my present-day avatar—a few months ago, we did so by training it on 20 years worth of books and interviews. We had a 4K video interview to train a video model, and we were able to use voice replication tools to replicate my voice. To create Reid AI 2004, my team had to get creative. It’s a custom GPT trained on knowledge up to 2004, including my own writing. Using Hedra, UNSHUT, and other tools, we relied largely on an image-to-video process that doesn’t require any training video but can generate video from AI-generated images. Overall, hearing myself pitch LinkedIn to myself was a bit surreal. I remember that early entrepreneur. This conversation between past and future selves feels not only like a reflection on how much has changed, but also on what remains constant: The core values of trust, integrity, and community that were pivotal in some of the startup's earliest pitches remain at the heart of LinkedIn today. I believe I would still invest in that idea. But I'm also amazed by how much 2004 Reid is about to see the world and technology change.  What do you think? What would you tell the younger you from a decade or two ago?

  • View organization page for Tavus, graphic

    8,112 followers

    One of the most interesting things about conversational AI video is how engaged people are talking to a digital twin and how surprised they are by how real it feels, even though they know it's AI. Today we launched on Hacker News, and the comments from developers are so thoughtful. Here is one of our favorites. Do you think talking to a digital twin knowingly will make you speak more humanly to it vs. the more transactional way we interact with Google via text? Read the comment for more context.

    • No alternative text description for this image
  • View organization page for Tavus, graphic

    8,112 followers

    👀 need weekend project inspo? How about building a personal website, with a digital twin. See how Hassaan Raza did it. ...or just talk to his digital twin.

    View profile for Hassaan Raza, graphic

    Co-Founder & CEO @ Tavus

    I've been wanting to build a personal website and landing page to video call with my Tavus digital twin for a while, and I finally did it. I used Replit and Cursor to do it, and man, was I blown away. This had been on my bucket list for a while- but a lack of time and my general displeasure in writing front-end had been blocking me. Using Replit was magic, letting me express my dreams and ideas and turning them into a reality. I was so impressed at its ability to plan things out, be verbose with its execution, and just rip through file by file—a complete step-change compared to copying and pasting code from Claude. Unfortunately, I hit my usage limit rather quickly (Amjad Masad, please increase the limits, friend), so then I exported the code and moved to Cursor. I’ve been using Copilot for a while and used Cursor to work on existing code, but I hadn’t used it or tested it for something from the ground up. Holy crap, I was incredibly impressed with it as well. I found that combining Anthropic Claude 3.5 Sonnet and OpenAI o1-preview worked well. It took a good degree of prompting to get it to do the right things at times, but o1-preview as an architect and Claude 3.5 sonner as the developer worked quite well. Highly recommend this setup. Also, throwing in o1-preview for harder problems or just to have an alternative take on something – was a game changer. Then, came actually integrating my Tavus digital twin into the site. I hooked up the Tavus docs and Cursor just ripped through and easily wrote code to create a conversation. While I may not have written much of the code for this personal blog, I did write all of the content because that’s the part that is truly me. AI cannot understand or replicate it. The best part of using tools like Replit and Cursor is that it allowed me to focus on writing the content and design rather than the intricacies of the front end. Before this, I didn’t think SWEs could be replaced, but now I think junior devs absolutely can be. I still do think it takes experience and software engineering knowledge to truly get great outputs. While creating anything may be easy, making something good will still take a sense of design and product think. That’s the human part that will not be replaced. Understanding system architecture and having an eye for design will be the differentiators. The point of this exercise was also to force me to put down my thoughts in a place for my digital twin persona to consume. This website will serve as the knowledge base for my digital twin. Creating my digital twin was super easy using Tavus. That will be the next post—how to create a digital twin of yourself easily using Tavus. For now, check out the website and talk to digital me. www.hassaanraza.com

  • View organization page for Tavus, graphic

    8,112 followers

    This week Meta promised avatars, digital twins (see demo with Don Allen Stevenson III), and more coming in 2025. This tech is already available at Tavus 😎. You can easily embed live digital twins into your platform using our APIs and end-to-end pipeline. We do everything from providing the LLM, TTS, Speech Recognition, Vision, and real-time Digital Twins. These layers are optimized to deliver the lowest latency on the market, at ~600ms between utterances (time between when a person speaks via video input, and the digital twin responds with a coherent, spoken, response on video). We build our own AI models, which offer a high degree of realism. If you're building apps or platforms in social media, marketing, sales, eCommerce, education, or other tools that need human-like video – reach out!

  • View organization page for Tavus, graphic

    8,112 followers

    Reid Hoffman, we love your digital twin experiments! If you're looking for real-time, you found it. The Tavus Conversational Video API 😉 . We have the lowest latency for real-time digital twins at ~600ms and getting faster with each day. Give us a call? Or you could always chat to our CEO Hassaan Raza's digital twin at https://lnkd.in/gEzta927.

    View profile for Reid Hoffman, graphic
    Reid Hoffman Reid Hoffman is an Influencer

    Co-Founder, LinkedIn & Inflection AI. Investor at Greylock.

    My digital twin—Reid AI—has been busy, from judging competitions to chatting with reporters to translating speeches. Some memorable moments from Reid AI's first few months of public appearances include: - Appearing in interviews with Andrew Ross Sorkin on CNBC and Bloomberg’s Emily Chang - Joining Allie K. Miller and me for an AI agent experiment - Helping me deliver my University of Perugia commencement speech in 10 languages - Discussing avatars' futures with actress / Simpson Street founder Kerry Washington, chef Massimo Bottura, and The Atlantic CEO Nicholas Thompson on the Possible podcast (https://lnkd.in/emENZKhA) - Judging a business plan competition with me at Stanford The list goes on... Aria Finger joined me to reflect on what Reid AI has been up to. So far, its superpowers include fluency in many languages and encyclopedic knowledge of my articles, books, and podcasts. Weaknesses: Not always responding intelligibly in real-time or giving the most nuanced advice (working on that...). I'm curious to continue exploring more applications of avatars, especially as they become increasingly easy to generate. I’m also starting to think there's a case to be made for Reid AI getting an executive assistant for all this work—AI, of course. What applications of avatars are you most excited about?

Similar pages

Browse jobs

Funding