Playtika Tech & AI’s Post

Our First Birthday! 🥳 Wow, what an incredible journey this first year has been! When we hit "publish" on the very first post a year ago, we had no idea the heights this little community would reach. From humble beginnings and a simple vision to share our love for tech and AI in gaming, we've grown into a global family of over 3,100 passionate members spanning 30+ countries! 🥳 milestone is all thanks to YOU - the amazing humans who showed up, engaged, contributed your creativity and smarts. 🥳You helped transform our nerdy passion project into a vibrant community. Whether you're a developer, engineer, AI wizard or just love this mind-blowing tech, we're so grateful you joined us on this wild ride. The journey has only just begun! 🔔 Afraid of missing out on our news? Worry not! Place the emoji 😀 in a comment below, and we'll make sure to tag you once we publish anything new. 👑 Here are our VIP alert subscribers: Vitalii Bortnik, Martyn Redstone, Aliaksei Nastsin, Guy Sebbag, Eugene Gladovskiy, Illia Shestakov, Albert Nicola, Samuel Wang, Cohi Elchadef Manobela, Adi Beker, Junwoo Yun, Maryanne Collins, Vakhtang Matskeplishvili, Odem Alagem אודם 🎡 , Marianna Inozemtseva, Alisa Yurchenko, Adrian Magearu.,Chen Offek, Evgenii Raisfeld, Kamil Sawko, Michal Lissak, Eze Vidra, Yash Baid, Cosmin Tanase, Shashikiran R, Raz Rosh, Almog Ben Simon

𝗠𝗮𝗺𝗯𝗮𝗕𝘆𝘁𝗲 𝗻𝗼𝘁 𝗠𝗮𝗺𝗯𝗮 𝗕𝗶𝘁𝗲 𝗠𝗮𝗺𝗯𝗮𝗕𝘆𝘁𝗲 is a new approach to language modeling that operates directly on 𝗯𝘆𝘁𝗲 𝘀𝗲𝗾𝘂𝗲𝗻𝗰𝗲𝘀 (individual units of digital information, typically representing characters in a more granular form) rather than 𝘁𝗼𝗸𝗲𝗻𝗶𝘇𝗲𝗱 𝘁𝗲𝘅𝘁 (text that has been broken down into larger units like words or subwords for easier processing by language models). 𝗦𝗼, 𝘄𝗵𝘆 𝗱𝗼 𝘄𝗲 𝗯𝘂𝘇𝘇 𝗮𝗯𝗼𝘂𝘁 𝘁𝗵𝗮𝘁: 🐍 MambaByte adapts the Mamba state space model architecture to work with raw byte sequences. 🐍 It achieves competitive performance with subword Transformers while being more robust to noise and typos. 🐍 The authors developed a speculative decoding technique using subword drafting and byte-level verification, achieving 2.6x faster inference. 🐍 Experiments show MambaByte outperforming byte-level Transformers and being competitive with subword models on language modeling tasks. 🐍 MambaByte appears to be more compute-efficient than alternatives like MegaByte Transformers. 🐍 The approach demonstrates the viability of using state space models for tokenizer-free language modeling at scale. 👉 The link to research paper is placed in comments below 👏 𝗣𝗹𝗲𝗮𝘀𝗲 𝗮𝗽𝗽𝗹𝗮𝘂𝗱 𝘁𝗵𝗲 𝗿𝗲𝘀𝗲𝗮𝗿𝗰𝗵𝗲𝗿𝘀 𝗶𝗻 𝘁𝗵𝗲 𝗰𝗼𝗺𝗺𝗲𝗻𝘁𝘀 𝗯𝗲𝗹𝗼𝘄 👏 Junxiong Wang, Tushaar Gangavarapu, Jing Nathan YAN, Alexander M. Rush 🐾 𝗢𝘂𝗿 𝗖𝗼𝗺𝗺𝗲𝗻𝘁 MambaByte ability to work directly with raw byte sequences could potentially simplify text processing pipelines. This technology could enable more sophisticated in-game language models and chatbots while reducing processing overhead. 🔔𝗔𝗳𝗿𝗮𝗶𝗱 𝗼𝗳 𝗺𝗶𝘀𝘀𝗶𝗻𝗴 𝗼𝘂𝘁 𝗼𝗻 𝗼𝘂𝗿 𝗻𝗲𝘄𝘀? Worry not! Place the emoji 😀 in a comment below, and we'll make sure to tag you once we publish anything new. 👑 Here are our VIP alert subscribers: Martyn Redstone, Maryanne Collins, Omer Gerson, Meta-Tech Stellar Systems, Aliaksei Nastsin, Illia Shestakov

𝗜𝗻𝘁𝗲𝗿𝗻 𝘁𝗵𝗮𝘁 𝗰𝗮𝗻 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗰𝗵𝗮𝗹𝗹𝗲𝗻𝗴𝗲 𝗚𝗣𝗧 𝗜𝗻𝘁𝗲𝗿𝗻𝗟𝗠 is a multilingual large language model with 104B parameters, jointly developed by Shanghai AI Laboratory and SenseTime 商汤科技 in collaboration with several universities. It was trained on a diverse corpus of 1.6T tokens, covering web text, books, academic papers, and code. InternLM has multilingual capabilities, with a focus on English and Chinese. It shows particular strength in Chinese language understanding and cultural knowledge. 𝗜𝗻𝘁𝗲𝗿𝗻𝗟𝗠 𝗼𝘂𝘁𝗽𝗲𝗿𝗳𝗼𝗿𝗺𝘀 𝗼𝘁𝗵𝗲𝗿 𝗼𝗽𝗲𝗻-𝘀𝗼𝘂𝗿𝗰𝗲 𝗺𝗼𝗱𝗲𝗹𝘀 𝗹𝗶𝗸𝗲 𝗚𝗟𝗠-𝟭𝟯𝟬𝗕 𝗮𝗻𝗱 𝗟𝗟𝗮𝗠𝗔-𝟲𝟱𝗕, 𝗮𝗻𝗱 𝗲𝘃𝗲𝗻 𝘀𝘂𝗿𝗽𝗮𝘀𝘀𝗲𝘀 𝗖𝗵𝗮𝘁𝗚𝗣𝗧 𝗼𝗻 𝗰𝗼𝗺𝗽𝗿𝗲𝗵𝗲𝗻𝘀𝗶𝘃𝗲 𝗲𝘅𝗮𝗺𝘀 𝗹𝗶𝗸𝗲 𝗠𝗠𝗟𝗨, 𝗔𝗚𝗜𝗘𝘃𝗮𝗹, 𝗖-𝗘𝘃𝗮𝗹, 𝗮𝗻𝗱 𝗚𝗔𝗢𝗞𝗔𝗢-𝗕𝗲𝗻𝗰𝗵. The model was developed using a multi-phase progressive pretraining approach and aligned using supervised fine-tuning, reward model training, and reinforcement learning from human feedback. Despite its impressive results, InternLM still faces limitations in areas such as very long article comprehension and complex reasoning, particularly when compared to GPT-4. 👉 Follow the link to know more https://lnkd.in/dQAGbppf 🐾 𝗢𝘂𝗿 𝗖𝗼𝗺𝗺𝗲𝗻𝘁 InternLM presents exciting opportunities for the mobile gaming industry. Such models could support greatly in-game dialogue systems, quest generation, and player support. They also have the potential to assist in game design and localization processes, potentially leading to more immersive and culturally nuanced gaming experiences across different markets. 🔔𝗔𝗳𝗿𝗮𝗶𝗱 𝗼𝗳 𝗺𝗶𝘀𝘀𝗶𝗻𝗴 𝗼𝘂𝘁 𝗼𝗻 𝗼𝘂𝗿 𝗻𝗲𝘄𝘀? Worry not! Place the emoji 😀 in a comment below, and we'll make sure to tag you once we publish anything new. 👑 Here are our VIP alert subscribers: Martyn Redstone, Maryanne Collins, Omer Gerson, Meta-Tech Stellar Systems

Meta 𝗰𝗮𝗻 𝗴𝗲𝗻𝗲𝗿𝗮𝘁𝗲 𝗮 𝘄𝗵𝗼𝗹𝗲 𝗭𝗼𝗼...𝗝𝘂𝘀𝘁 𝘁𝗲𝘅𝘁 𝗶𝘁! AI at Meta has introduced 𝟯𝗗𝗚𝗲𝗻, a pipeline for text-to-3D asset generation. This system can create high-quality 3D assets with good prompt fidelity in under a minute. 3DGen supports physically-based rendering (PBR) and can generate or retexture 3D shapes using textual inputs. The pipeline combines two main components: - 𝗠𝗲𝘁𝗮 𝟯𝗗 𝗔𝘀𝘀𝗲𝘁𝗚𝗲𝗻 for text-to-3D generation and - 𝗠𝗲𝘁𝗮 𝟯𝗗 𝗧𝗲𝘅𝘁𝘂𝗿𝗲𝗚𝗲𝗻 for text-to-texture generation. This allows 3DGen to represent 3D objects in view space, volumetric space, and UV (texture) space simultaneously. In user studies, 3DGen performed well compared to several industry baselines, including Tripo3D, Rodin Gen-1, Meshy v3, and other third-party text-to-3D generators. It showed strong performance in prompt fidelity, overall visual quality, texture quality, and geometry accuracy. 3DGen's advantage becomes more noticeable as prompt complexity increases, particularly for characters and compositions. 👉 Follow the link to know more https://lnkd.in/dkHUkmwv 👏 Please applaud the researchers in the comments below 👏 Raphael Bensadoun, Tom Monnier, Yanir Kleiman, Filippos Kokkinos, Yawar Siddiqui, Mahendra K., Omri Harosh, Roman S., Benjamin Graham, Emilien Garreau, Animesh Karnewar, Ang Cao, Idan Azuri, Iurii Makarov, Eric-Tuan Lê, Antoine Toisoul, David Novotný, Oran Gafni, Natalia Neverova, Andrea Vedaldi 🐾 𝗢𝘂𝗿 𝗖𝗼𝗺𝗺𝗲𝗻𝘁 This technology could potentially allow players to create and customize their own 3D elements through text descriptions, which could enhance engagement and creativity within our gaming ecosystem. 🔔𝗔𝗳𝗿𝗮𝗶𝗱 𝗼𝗳 𝗺𝗶𝘀𝘀𝗶𝗻𝗴 𝗼𝘂𝘁 𝗼𝗻 𝗼𝘂𝗿 𝗻𝗲𝘄𝘀? Worry not! Place the emoji 😀 in a comment below, and we'll make sure to tag you once we publish anything new. 👑 Here are our VIP alert subscribers: Martyn Redstone, Maryanne Collins, Michal Lissak, Omer Gerson, Adrian Magearu, Shai Oliva Alon

𝗠𝗲𝗮𝘀𝘂𝗿𝗲 𝘁𝗵𝗮𝘁! Anthropic is launching a program to fund the development of new types of benchmarks to evaluate the performance and impact of AI models, including generative models like Claude. The program will provide funding to third-party organizations that can effectively measure advanced capabilities in AI models. Anthropic aims to create challenging benchmarks focused on AI security and societal implications, using new tools, infrastructure and methods. Specific areas of focus include: - assessing models' ability to carry out cyberattacks, - enhance weapons of mass destruction, - manipulate people, - aid scientific study, - converse in multiple languages, - mitigate biases, and self-censor toxicity. Anthropic envisions new platforms allowing experts to develop evaluations and large-scale trials involving thousands of users. The company has hired a full-time coordinator and may purchase or expand promising projects. Funding options will be tailored to each project's needs and stage. Anthropic wants some evaluations to align with its own AI safety classifications. The program aims to catalyze progress towards comprehensive AI evaluation as an industry standard. Some experts may criticize Anthropic's focus on "catastrophic" AI risks as distracting from more immediate regulatory issues. 👉 Follow the link to know more https://lnkd.in/eRT63kt9 🐾 𝗢𝘂𝗿 𝗖𝗼𝗺𝗺𝗲𝗻𝘁 Three months ago, we identified a gap in AI benchmarking within the mobile gaming sector. 𝗧𝗵𝗲𝗿𝗲 𝗶𝘀 𝗮𝗻 𝘂𝗿𝗴𝗲𝗻𝘁 𝗻𝗲𝗲𝗱 𝗳𝗼𝗿 𝗺𝗼𝗿𝗲 𝗱𝗶𝘃𝗲𝗿𝘀𝗲 𝗮𝗻𝗱 𝘀𝗼𝗽𝗵𝗶𝘀𝘁𝗶𝗰𝗮𝘁𝗲𝗱 𝗯𝗲𝗻𝗰𝗵𝗺𝗮𝗿𝗸𝘀 𝘁𝗵𝗮𝘁 𝗮𝗰𝗰𝘂𝗿𝗮𝘁𝗲𝗹𝘆 𝗲𝘃𝗮𝗹𝘂𝗮𝘁𝗲 𝗔𝗜 𝗽𝗲𝗿𝗳𝗼𝗿𝗺𝗮𝗻𝗰𝗲 𝗶𝗻 𝘁𝗵𝗶𝘀 𝘂𝗻𝗶𝗾𝘂𝗲 𝗰𝗼𝗻𝘁𝗲𝘅𝘁. 👉 We invite you to read analysis on why existing benchmarks fall short in providing accurate assessments and how the industry can address this issue: https://lnkd.in/divE-YXN 🔔𝗔𝗳𝗿𝗮𝗶𝗱 𝗼𝗳 𝗺𝗶𝘀𝘀𝗶𝗻𝗴 𝗼𝘂𝘁 𝗼𝗻 𝗼𝘂𝗿 𝗻𝗲𝘄𝘀? Worry not! Place the emoji 😀 in a comment below, and we'll make sure to tag you once we publish anything new. 👑 Here are our VIP alert subscribers: Martyn Redstone, Illia Shestakov, Albert Nicola, Maryanne Collins, Vakhtang Matskeplishvili, Michal Lissak, Omer Gerson

𝗧𝗲𝗮𝗰𝗵𝗲𝗿𝘀 𝗶𝗻 𝟭𝟵𝟵𝟬: 𝗟𝗲𝗼𝗻𝗮𝗿𝗱𝗼 𝗶𝘀 𝗻𝗼𝘁 𝗮 𝘁𝘂𝗿𝘁𝗹𝗲! 𝗧𝗲𝗮𝗰𝗵𝗲𝗿𝘀 𝗶𝗻 𝟮𝟬𝟮𝟰: 𝗟𝗲𝗼𝗻𝗮𝗿𝗱𝗼 𝗱𝗮 𝗩𝗶𝗻𝗰𝗶 𝗶𝘀 𝗻𝗼𝘁 𝗔𝗜. 𝗔𝗻𝗱 𝗶𝘁’𝘀 𝗻𝗼𝘁 𝗮 𝘁𝘂𝗿𝘁𝗹𝗲 𝗲𝗶𝘁𝗵𝗲𝗿! Leonardo.Ai has launched a new image generation model, challenging Ideogram's position as one of the leading AI image generators. Both companies are competing to provide high-quality text-to-image capabilities, with a focus on 𝘁𝗲𝘅𝘁 𝗿𝗲𝗻𝗱𝗲𝗿𝗶𝗻𝗴 𝗮𝗻𝗱 𝗹𝗼𝗴𝗼 𝗰𝗿𝗲𝗮𝘁𝗶𝗼𝗻. Follow the link to know more https://lnkd.in/daerzg28 🐾 𝗢𝘂𝗿 𝗖𝗼𝗺𝗺𝗲𝗻𝘁 Tools like Leonardo AI and Ideogram are especially valuable for conceptualizing and brainstorming. 👉 𝗪𝗲 𝘂𝘀𝗲𝗱 𝘁𝗵𝗲 𝘀𝗮𝗺𝗲 𝗽𝗿𝗼𝗺𝗽𝘁 𝗶𝗻 𝗯𝗼𝘁𝗵 𝘁𝗼𝗼𝗹𝘀. 👉 𝗪𝗵𝗶𝗰𝗵 𝗿𝗲𝘀𝘂𝗹𝘁 𝗱𝗼 𝘆𝗼𝘂 𝗹𝗶𝗸𝗲 𝘁𝗵𝗲 𝗺𝗼𝘀𝘁? 👉 See the comments and vote 👇 🔔𝗔𝗳𝗿𝗮𝗶𝗱 𝗼𝗳 𝗺𝗶𝘀𝘀𝗶𝗻𝗴 𝗼𝘂𝘁 𝗼𝗻 𝗼𝘂𝗿 𝗻𝗲𝘄𝘀? Worry not! Place the emoji 😀 in a comment below, and we'll make sure to tag you once we publish anything new. 👑 Here are our VIP alert subscribers: Martyn Redstone, Illia Shestakov, Albert Nicola, Maryanne Collins, Michal Lissak, Omer Gerson, Adrian Magearu, Shai Oliva Alon, Asya Polyak

𝗔𝘀 𝗔𝗜 𝗮𝗱𝘃𝗮𝗻𝗰𝗲𝘀, 𝗰𝗼𝘀𝘁𝘀 𝘀𝗼𝗮𝗿. 𝗪𝗵𝗮𝘁 𝗰𝗮𝗻 𝘄𝗲 𝗱𝗼 𝗮𝗯𝗼𝘂𝘁 𝘁𝗵𝗮𝘁? Etched is making an ambitious bet on specialized transformer AI chips. This 𝗔𝗦𝗜𝗖 (Application-Specific Integrated Circuit) is designed exclusively for transformer models, like 𝗖𝗵𝗮𝘁𝗚𝗣𝗧, 𝗣𝗮𝗟𝗠, and 𝗟𝗟𝗮𝗠𝗮. 𝗦𝗼𝗵𝘂 is claimed to be significantly faster and cheaper than NVIDIA's next-gen 𝗕𝗹𝗮𝗰𝗸𝘄𝗲𝗹𝗹 GPUs, offering over 500,000 tokens per second for Llama 70B. The bet is based on the belief that scale is crucial for AI advancement. Major players like OpenAI, Anthropic, Meta, and Google are doubling down on scaling transformers, with training runs expected to cost billions or even trillions of dollars. 𝗘𝘁𝗰𝗵𝗲𝗱 believes that specializing hardware for this dominant architecture is inevitable at such scales, much like bitcoin mining ASICs disrupted GPUs. While a risky bet if transformer architectures are supplanted, Etched argues that the transformer's "moat" - optimized software stacks, massive R&D investments, and the astronomical costs of re-validating new architectures - makes a disruption unlikely in the near term. If transformers persist, 𝗦𝗼𝗵𝘂 could revolutionize AI capabilities by making real-time video generation, coding assistants, and other compute-intensive tasks practical and affordable. 👉 Follow the link to know more https://lnkd.in/dYabYHcc 🐾 𝗢𝘂𝗿 𝗖𝗼𝗺𝗺𝗲𝗻𝘁 The potential of Etched's Sohu chip is intriguing.  Transformer models could open new frontiers for intelligent agents, procedural content generation, and hyper-personalized experiences within games. However, it is worth to consider the challenges of integrating such powerful hardware into mobile devices and the potential need for cloud-based solutions to leverage this technology effectively in mobile games. 🔔𝗔𝗳𝗿𝗮𝗶𝗱 𝗼𝗳 𝗺𝗶𝘀𝘀𝗶𝗻𝗴 𝗼𝘂𝘁 𝗼𝗻 𝗼𝘂𝗿 𝗻𝗲𝘄𝘀? Worry not! Place the emoji 😀 in a comment below, and we'll make sure to tag you once we publish anything new. 👑 Here are our VIP alert subscribers: Martyn Redstone, Illia Shestakov, Albert Nicola, Maryanne Collins, Vakhtang Matskeplishvili, Michal Lissak, Omer Gerson

𝗪𝗮𝗻𝘁 𝘁𝗼 𝗹𝗲𝗮𝗿𝗻 𝗳𝗿𝗼𝗺 𝘁𝗵𝗲 𝗯𝗲𝘀𝘁? Then get ready for an exciting journey with Oded Nadel as your guide! Playtika AI Product Manager who took 𝗣𝗹𝗮𝘆𝘁𝗶𝗸𝗮'𝘀 𝗠𝗟 𝗽𝗹𝗮𝘁𝗳𝗼𝗿𝗺 from idea to reality. In the first part, Oded opens up about the challenges and constraints they faced. But he also shares how they analyzed the situation and made wise choices. This is a rare opportunity to go behind the scenes and witness how Playtika's top engineers conjured up the 𝗠𝗟𝗼𝗱𝘆 𝗽𝗹𝗮𝘁𝗳𝗼𝗿𝗺 entirely from scratch. It was a wild ride, but Oded explains each step in detail. 🐾 Follow the link 👉 https://lnkd.in/dCxCrA5K 👈 🔔𝗔𝗳𝗿𝗮𝗶𝗱 𝗼𝗳 𝗺𝗶𝘀𝘀𝗶𝗻𝗴 𝗼𝘂𝘁 𝗼𝗻 𝗼𝘂𝗿 𝗻𝗲𝘄𝘀? Worry not! Place the emoji 😀 in a comment below, and we'll make sure to tag you once we publish anything new. 👑 Here are our VIP alert subscribers: Martyn Redstone, Illia Shestakov, Albert Nicola, Maryanne Collins, Omer Gerson

𝗡𝗼𝘁 𝗷𝘂𝘀𝘁 𝗮 𝘁𝗮𝗹𝗸𝗶𝗻𝗴 𝗵𝗲𝗮𝗱 Synthesia 2.0 is an 𝗔𝗜 𝘃𝗶𝗱𝗲𝗼 𝗰𝗼𝗺𝗺𝘂𝗻𝗶𝗰𝗮𝘁𝗶𝗼𝗻𝘀 𝗽𝗹𝗮𝘁𝗳𝗼𝗿𝗺 that enables users to create engaging videos from text. It introduces several new features, including 𝗘𝘅𝗽𝗿𝗲𝘀𝘀𝗶𝘃𝗲 𝗔𝘃𝗮𝘁𝗮𝗿𝘀 that can display emotions and change tone based on the script context. The 𝗠𝘂𝗹𝘁𝗶𝗹𝗶𝗻𝗴𝘂𝗮𝗹 𝗩𝗶𝗱𝗲𝗼 𝗣𝗹𝗮𝘆𝗲𝗿 automatically plays videos in the viewer's language. 𝗟𝗶𝘃𝗲 𝗖𝗼𝗹𝗹𝗮𝗯𝗼𝗿𝗮𝘁𝗶𝗼𝗻 allows inviting colleagues, real-time collaboration, commenting, and faster workflow. The 𝗔𝗜 𝗩𝗶𝗱𝗲𝗼 𝗔𝘀𝘀𝗶𝘀𝘁𝗮𝗻𝘁 can convert documents, links, or ideas into brand-consistent videos. 𝗗𝗶𝗮𝗹𝗼𝗴𝘂𝗲 enables including multiple avatars in scenes to recreate engaging scenarios. The 𝗔𝗜 𝗦𝗰𝗿𝗲𝗲𝗻 𝗥𝗲𝗰𝗼𝗿𝗱𝗲𝗿 transcribes voiceovers, syncs them with screen captures, and adds automatic zoom effects. Looking ahead, Synthesia plans to introduce personalized digital twins with custom backgrounds, interactive video stories, and next-generation AI avatars with full-body motions.  The company is also pursuing ISO 42001 certification, demonstrating its commitment to responsible AI use. 👉 We have attached a part of demonstration video for your convenience 👉 Follow the link to see full demonstration video https://lnkd.in/df5mfiy5 👉 Follow the link to know more https://www.synthesia.io/2 🐾 𝗢𝘂𝗿 𝗖𝗼𝗺𝗺𝗲𝗻𝘁 Synthesia might work great for crafting game tutorials and marketing materials. Something that was takien day now can be done in hours. 🔔𝗔𝗳𝗿𝗮𝗶𝗱 𝗼𝗳 𝗺𝗶𝘀𝘀𝗶𝗻𝗴 𝗼𝘂𝘁 𝗼𝗻 𝗼𝘂𝗿 𝗻𝗲𝘄𝘀? Worry not! Place the emoji 😀 in a comment below, and we'll make sure to tag you once we publish anything new. 👑 Here are our VIP alert subscribers: Martyn Redstone, Illia Shestakov, Albert Nicola, Maryanne Collins, Omer Gerson, Marianna Inozemtseva, Adrian Magearu

𝗢𝗵! 𝗥𝗲𝗹𝗶𝗮𝗯𝗹𝗲 𝗮𝗴𝗲𝗻𝘁 𝗱𝗲𝗽𝗹𝗼𝘆! 𝗠𝗲 𝘄𝗮𝗻𝗻𝗮! LangChain has announced the stable release of 𝗟𝗮𝗻𝗴𝗚𝗿𝗮𝗽𝗵 𝘃𝟬.𝟭, a framework for building agentic and multi-agent applications with better control and precision. Companies like Klarna, Replit, Ally, Elastic, and NCL are already utilizing LangGraph for their AI initiatives. Alongside LangGraph v0.1, LangChain introduced 𝗟𝗮𝗻𝗴𝗚𝗿𝗮𝗽𝗵 𝗖𝗹𝗼𝘂𝗱, currently in closed beta. 𝗟𝗮𝗻𝗴𝗚𝗿𝗮𝗽𝗵 𝗖𝗹𝗼𝘂𝗱 is a scalable infrastructure for deploying LangGraph agents reliably. It handles task queues, servers, checkpointing, double-texting strategies, async jobs, and cron jobs. It integrates with 𝗟𝗮𝗻𝗴𝗚𝗿𝗮𝗽𝗵 𝗦𝘁𝘂𝗱𝗶𝗼 for visualization and 𝗟𝗮𝗻𝗴𝗦𝗺𝗶𝘁𝗵 for deployment and monitoring. LangGraph's flexible API enables developers to design custom cognitive architectures with low-level control over code flow, prompts, and LLM calls. 👉 Follow the link for more https://lnkd.in/gdBGWkVs 👉 GitHub project https://lnkd.in/drYHQhFx 👉 Any questions? Ask Mike Bird in the comments bellow 🐾 𝗢𝘂𝗿 𝗖𝗼𝗺𝗺𝗲𝗻𝘁 LangGraph v0.1 and LangGraph Cloud can be great tools for creating intelligent NPCs, designing adaptive gameplay experiences, and optimizing user engagement. 🔔𝗔𝗳𝗿𝗮𝗶𝗱 𝗼𝗳 𝗺𝗶𝘀𝘀𝗶𝗻𝗴 𝗼𝘂𝘁 𝗼𝗻 𝗼𝘂𝗿 𝗻𝗲𝘄𝘀? Worry not! Place the emoji 😀 in a comment below, and we'll make sure to tag you once we publish anything new. 👑 Here are our VIP alert subscribers:  Martyn Redstone, Illia Shestakov, Albert Nicola, Maryanne Collins, Vakhtang Matskeplishvili, Kamil Sawko, Omer Gerson