news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

Today's Financing Express

AI image generation platform Liblib AI raises hundreds of millions of yuan in financing

LiblibAI, an AI image generation platform, has completed three rounds of financing within a year, with a total amount of hundreds of millions of RMB. The angel investors are Source Code Capital, Gaorong Ventures and GSR Ventures; the second round was led by strategic investors; the third round was led by Vision Capital; and old shareholders continued to support the company for multiple rounds. Among them, Farsight Capital is the exclusive financial advisor for multiple rounds of financing. (Intelligence Emerges)

Pet empathy AI company Traini receives tens of millions of angel round financing

The AI ​​application Traini not only allows users to understand the dog's barking, but also understand the dog's facial expressions, body behavior and other languages ​​to understand the dog's psychological condition. Traini is a pet empathy AI application founded by Chinese serial entrepreneur Sun Linjia in Silicon Valley, USA. In the case of cold start, it has 200,000 registered iOS users. In May this year, the product has been viewed more than one million times.

Traini completed a 10 million yuan angel round of financing in May, with investors including FutureX Capital, BlueSea Partners, Tao Foundation, Mint Capital and other investors, including former Meta engineering director Liang Yingyi, Caper AI (Instacart subsidiary) engineering director Lily Zhang, ORH Partners partner Jia Peilin, blogger Janie Tanmei, etc. Old shareholders Valkyrie Fund I LP and Fengmin Gong (Palo Alto Networks Cofounder) continued to invest.

AI-driven blood cell analysis platform Scopio Labs receives $42 million in Series D funding

Scopio Labs' platform uses computational photography technology to provide digital imaging of thousands of cells, providing laboratory practitioners with a high-resolution full-scene view, eliminating the need for manual microscope operation. The company's solution combines AI-supported clinical decision-making and full-scene digital imaging to improve the efficiency of blood cell analysis workflows, promote remote consultation and collaboration, and speed up decision-making, thereby improving patient care.

Scopio Labs, the developer of the Full-Field Digital Cell Morphology imaging and analysis platform, recently announced the completion of a $42 million Series D round of financing. The round was led by Fortissimo Capital, with participation from other investors including OurCrowd, Aurum Ventures MKI, OTV, and LR Group.

AI chip company Fractile receives $15 million in seed funding

Founded in 2022 by 28-year-old Walter Goodwin, a PhD in artificial intelligence, Fractile is committed to building the first new AI chip capable of running state-of-the-art AI models 100 times faster and 10 times cheaper than existing hardware.

Kindred Capital, NATO Innovation Fund and Oxford Science Enterprises led the round, with participation from Cocoa and Inovia Capital, and angel investors including Hermann Hauser (co-founder of Acorn and Amadeus Capital), Stan Boland (formerly of Icera, NVIDIA, Element 14 and Five AI), and Amar Shah (co-founder of Wayve). To date, Fractile has raised a total of $17.5 million.

AI-driven data platform DAO Base completes $6 million in financing

DAO Base is an AI-driven data and infrastructure platform designed specifically for DAOs. It helps DAOs and voters build their blockchain and non-blockchain data reputation layer through advanced algorithms, enabling users to intelligently monitor DAO governance, make informed decisions, and explore future working models.

DAOBase announced that it has successfully raised $6 million in funding. This round of financing was supported by multiple investors including SevenX Ventures, AntAlpha Ventures (Bitmain), y2z Ventures, Tanuki Studio, Slingshot Labs, LD Capital, Mix Marvel DAO Venture, TI Capital and Bitrise Capital, as well as participation from individual angel investors such as The Sandbox co-founder Sébastien Borget and DeXe Protocol co-founder Dmytro Kotliarov.

AI-powered storytelling platformLumi StoryReceived $4 million in seed financing

Founded by former NFL quarterback Colin Kaepernick, Lumi is an AI-driven storytelling platform designed to help creators develop, publish and distribute their stories. The platform provides tools for creating story characters, planning narratives, and generating scripts. Lumi's AI assists with visual and text elements, helping to produce graphic novels and other content. The platform also provides publishing and sales services and logistics processing so that creators can focus on their creations. This round of financing was led by Seven Seven Six.

Gen AI startup Rabbitt.AI raises $2.1 million

Rabbitt.AI was founded by IIT-D alumni and recently received $2.1 million in funding from TC Group of Companies. The company provides customized LLM development tools, RAG fine-tuning, and creation and deployment services for advanced AI applications such as data center AI. The platform has MLOps sets, voice robot AI agents, etc., and prioritizes private-first strategies in AI deployment.

“Smaller, custom, and industry-specific fine-tuned models are making more progress than one large model,” said Harneet SN, founder and chief officer of Rabbitt.AI. “At Rabbitt.AI, we are advocates of these models and help companies adopt open source AI models.”

(Welcome to add WeChatAIyanxishe2, learn more about AIGC and financing, and chat with like-minded friends about new AI products)

Today's big factory news

Meta training Llama 3 encounters frequent failures: 16384 H100 GPU training cluster "strikes" every 3 hours

A research report released by Meta showed that its 16,384 Nvidia H100 graphics card clusters used to train the 405 billion parameter model Llama 3 experienced 419 unexpected failures in 54 days, an average of one every three hours. More than half of the failures were caused by the graphics card or its high-bandwidth memory (HBM3). GPU problems were the main cause of failures, accounting for 58.7% of unexpected outages. (IT Home)

Baichuan Intelligence established an information service company with a registered capital of 50 million yuan

Beijing Baichuan Intelligent Information Service Co., Ltd. was recently established, with Deng Jiang as the legal representative and a registered capital of 50 million yuan. Its business scope includes software development, computer system services, artificial intelligence application software development, artificial intelligence entrepreneurship and innovation service platform, information technology consulting services, big data services, etc. According to shareholder information, the company is wholly owned by Beijing Baichuan Intelligent Technology Co., Ltd., which is controlled by Wang Xiaochuan.

Anyone can permanently access the deleted GitHub code, GitHub: It's not a bug

Researchers at Truffle Security found that even if the code repository is deleted on GitHub, the data of the code and its forked copies can still be accessed. GitHub's parent company Microsoft responded that this is an intentional "feature" and not a bug. GitHub said that this design is consistent with the description in the official documentation and the effect is exactly as expected. Truffle Security CEO Dylan Ayrey suggested that GitHub consider not sharing the fork pool between forks and establish a new function to allow users to permanently delete commits.

Wave Intelligence releases the next generation of personalized adaptive private language model "Weaver 2.0"

Waveform Intelligence has released a new generation of multi-modal, infinitely long content generation, personalized adaptive private language model "Weaver 2.0", as well as upgraded multi-language and multi-modal AI content creation tools "WaWa Writing 2.0", "Siuuu.AI", "AI Learning" and other products. Currently, the new generation of "Weaver 2.0" model and LPA technology have been deployed in multiple products and applications of Waveform Intelligence.

Jiang Yuchen, founder and CEO of Waveform Intelligence, graduated from the Zhejiang University Zhu Kezhen College with a bachelor's degree and obtained a doctorate from the Swiss Federal Institute of Technology. He studied under the internationally renowned scholar Ryan Cotterell and worked on the training and reasoning of large language models at Microsoft Research Asia. In January this year, Waveform Intelligence completed a Pre-A round of financing worth tens of millions of yuan, led by BlueRun Ventures, followed by Westlake Science and Technology Ventures, Ant Financial Chairman Jing Xiandong, well-known angel investor Fei Gesheng, and old shareholder Oufang Angel. (36Kr)

38,500 yuan, all open source! Zhongqing Robotics' first EDU version of professional-grade adult doll is now on sale

Zhongqing Robotics has launched the SA01 EDU version of a professional-grade humanoid robot, targeting the scientific research and education market. The product is available as a complete machine or in a kit of parts, priced at RMB 38,500. The robot is highly scalable and can be equipped with a robotic arm and other humanoid parts. SA01 uses a reinforcement learning algorithm and a high-efficiency power module solution, with excellent power performance and a high-quality aluminum alloy skeleton, ensuring long life and high cost performance. Zhongqing has completed the angel round of financing and is preparing for a new round of financing. In addition, two full-size humanoid robot products will be launched in the near future, targeting industrial and commercial service scenarios respectively.

Kaiying Network releases new breakthrough in AI big model, making fully automatic game generation a reality

Kaiying Network announced a breakthrough in AI technology in game production, namely how to achieve fully automatic game generation through the application of large models and realize a change in the way games are produced. The advanced multi-task execution released this time makes AI generation not limited to text, audio and video content, but a composite tool that can meet the development needs of a game in multiple dimensions. Based on this development logic, Kaiying's large model provides developers with a question-and-answer application, guiding users to make demands on AI in the form of "chatting" with people. AI will automatically start drawing, writing documents, making art, making music, and then automatically integrating a complete game according to the needs, which makes the threshold for making a game almost zero.

todayProduct News

Product HuntHot List, AI Product Architect Archie

Archie is an AI-driven product architect that compresses traditional months of work in minutes and significantly reduces costs. Archie leverages multiple large language models to specialize in different stages of the product lifecycle. Allowing users to focus on creative work rather than prompt engineering, it enables a faster and smarter design and planning process through efficient AI API calls. Archie can help users think more comprehensively, identify potential problems in advance, and provide suggestions for improvements, such as compliance, security and other considerations. Archie's workflow includes stages such as creativity, requirements, visual architecture, technical architecture, and planning, and each stage can save time and costs through intelligent processing of AI. Users can ask questions, make requests, and even have fun with the project through the chat function with Archie.

https://archie.8base.com/?ref=ph_home&utm_campaign=Archie&utm_medium=webdisplay&utm_source=producthunt

GitHub Trending Hot List, open source project ollama

ollama is an open source project designed to help users easily start and run large language models including Llama 3.1, Mistral, Gemma 2. The project supports multiple operating systems, including macOS, Windows, and Linux, and provides a Docker image ollama. Users can download, create, run, and manage models through simple command line operations. The project also provides a series of libraries such as ollama-python and ollama-js, as well as detailed documentation and examples to help developers integrate and customize models. In addition, ollama supports importing models from GGUF, PyTorch, or Safetensors formats, and allows users to customize model parameters and prompts by writing Modelfile. The installation and usage instructions of this project are very clear and suitable for users of all technical levels.

https://github.com/ollama/ollama

Developer Recommendation

1. Alchemist: Changing object properties in real images

MIT CSAIL and Google Research team introduced a technology called "Alchemist", which can flexibly edit the material properties of objects in real images without losing the authenticity of the image. This technology supports linear adjustment of material properties such as roughness, transparency, metalness and albedo of objects.

https://prafullsharma.net/alchemist/

2. Differ: Insert objects based on text description

Xiamen University, OpenGVLab, and the University of Hong Kong proposed Diffree, which can add objects naturally through text descriptions without the need for users to specify bounding boxes or hand-drawn masks. By training on the OABench dataset, Diffree can handle a variety of object addition tasks and can insert objects multiple times in the same image. A key feature of Diffree is that it can predict the location of new objects and can add objects through text guidance.

Project https://opengvlab.github.io/Diffree/

试玩https://huggingface.co/spaces/LiruiZhao/Diffree

Special attention

Johnny Ho, co-founder and head of product at Perplexity, reveals how to build a company with AI

Perplexity is an AI company that was founded less than two years ago, with less than 50 employees, tens of millions of users, and annual revenue of more than $20 million. Recently, Perplexity completed a $63 million financing, with a valuation of more than $1 billion, attracting the attention of well-known investors including Nvidia and Jeff Bezos.

Perplexity's product development approach includes using AI for product management, a slime mold-like organizational structure, small teams, a small number of managers, and predictions about the future value of product managers. Employees are encouraged to solve problems through AI first. Perplexity's team is usually only two or three people, and some projects, such as podcasts, are even completed by one person alone. The company tends to hire self-driven individuals rather than those who emphasize management skills.

Perplexity uses AI tools to build its products, and despite limited application in coding, it is very effective in product management and decision making. The company has only two full-time product managers, and projects are usually completed by one or two people, with no more than four people. The main work of the product manager is to judge and prioritize use cases, as well as make decisions based on data and user research.

Perplexity's team structure is designed to minimize coordination costs and parallelize projects by sharing reusable guidelines and processes. The company uses AI for "rubber duck debugging" to reduce reliance on perfect consistency and consensus. Perplexity has quarterly plans and maintains flexibility in its product roadmap to adapt to the rapid changes in the AI ​​field. The company's decision-making process is relatively decentralized, projects are driven by a single responsible person, and execution steps are parallelized as much as possible. Perplexity's product development success depends on distilling feedback into intuitive products and giving team members the power to control decisions for rapid iteration.

https://www.lennysnewsletter.com/p/how-perplexity-builds-product

Please stay tuned for the latest updates tomorrow！

AI Intelligence Agency is looking for intelligence partners to collect exclusive valuable clues! If you can provide information about the latest AI achievements, industry insider information, and unique products, please add the operation WeChat account:AIyanxishe2Note the industry position.

Zhipu received a strategic investment of RMB 100 million from Huace to build a film and television AI research platform; Yang Hongxia joined the Hong Kong Polytechnic University; NIO released China's first intelligent driving world model | AI Intelligence Bureau

Baichuan Intelligence completes 5 billion yuan Series A financing, with state-owned capital from Beijing, Shanghai and Shenzhen entering the market; OpenAI challenges Google and releases AI search SearchGPT丨AI Intelligence Bureau

AI legal company Clio receives $900 million in financing; Alibaba Cloud and Tencent Cloud launch Llama 3.1 training and reasoning services; KeLing international version is launched | AI Technology Review