news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

Today's Financing Express

AI-based cloud platform osapiens receives $120 million in Series B funding

osapiens is an ESG software solution provider that develops software-as-a-service solutions to help companies across industries implement ESG requirements quickly, automatically, and securely. The core of the company is osapiens HUB, an AI-based cloud platform that creates compliance and transparency across the value chain. It allows companies to address all ESG challenges to identify risks, implement reporting requirements such as CSRD, EUDR, and CSDDD, and make their operational processes more sustainable in the long term. Recently, osapiens received $120 million in Series B financing from Goldman Sachs Alternatives. (IT Orange)

AI basic software developer Chuangjushou Technology received 20 million yuan in angel round financing

Chuangjushou Technology is a digital marketing system service provider. The company is mainly engaged in software development, software sales, artificial intelligence basic software development, pay-before-pay system development and other related businesses. Recently, Chuangjushou completed an angel round of financing of RMB 20 million. This round of financing was invested by investment institution Fuxing Capital. The funds will be mainly used for the further development and technology upgrade of the company's pay-before-pay system.

AI accounting platform Rillet receives $13.5 million in funding

Rillet announced it has raised $13.5 million to build a next-generation accounting platform. The company uses metadata and AI to understand raw source data and can automate a variety of workflows that finance teams used to have to do manually — from invoicing to closing accounts and running investor reports.

The financing was led by Creandum and First Round Capital, with participation from individual investors including Chad Byers (Susa Ventures), Kevin Hartz (founder of Eventbrite and Xoom), the former chief accounting officer of Facebook and Stripe, and the CFO of Ramp.

Wearable AI assistant Bee AI raises $7 million

Bee AI has raised $7 million in a round led by Exor to develop a wearable AI assistant that listens to you, learns more about you, takes notes, displays contextual reminders, and builds lists. The company is also launching a companion Apple Watch app. Greycroft, New Wave VC, Banana Capital, and Brian Bedol also participated in the new round.

AI-driven recruitment intelligence platform Tech1M receives $750,000 in seed round funding

Tech1M, a recruitment intelligence platform provider, mainly provides AI-driven recruitment automation and skills assessment software services that enable companies to quickly attract, evaluate and hire candidates. Tech1M raised $750,000 in seed funding led by SFC Capital (which recently backed 4D Medicine and VoltShare) and Techstar. Tech Nation, Barclays Eagle Labs, HSBC Innovation Banking, ICE Network, Vantage, 1000 Black Voices, Foundervine and GrowLondon Global also participated in this round of financing.

Flink Robotics raises €156,000 to transform industrial robots into dynamic workers

Industrial automation startup Flink Robotics receives €156,000 in funding from Venture Kick to enable robots to adapt to dynamic environments. Flink Robotics, which transforms robots into adaptable, plug-and-play workers for a wide range of dynamic material handling tasks, received funding from Venture Kick.

Canva acquires Leonardo.ai to bolster its generative AI efforts

Canva has acquired Leonardo.ai. The financial terms of the deal were not disclosed, but Canva co-founder and chief product officer Cameron Adams said it was a mix of cash and stock. All of Leonardo.ai’s 120 employees will join Canva, including the management team.

Sydney-based Leonardo.ai was founded in 2022 and was originally designed to focus on video game asset creation. The team at Leonardo.ai then decided to build the platform to meet more scenarios, such as creating and training AI models for image creation in industries such as fashion, advertising, and architecture. Today, Leonardo.ai provides collaborative tools and a private cloud for models including video generators, and provides API access so that customers can build their own technical infrastructure on Leonardo.ai's platform.

(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

Apple launches AI version of iOS, using Google chips, introducing new Siri and writing features

Apple's latest Apple Intelligence was launched with iOS 18.1 Beta, bringing users features including a comprehensive Siri upgrade, writing assistance, and text generation. Siri not only has an upgraded interface, but also supports text conversations and can better understand context and answer questions.

Apple's own large model, the Apple Foundation Model (AFM), has both client-side and cloud-side versions, and surpasses GPT-4 on multiple tasks. AFM is trained using a Google TPU cluster rather than Nvidia hardware, and uses a unique reinforcement learning algorithm and client-side mixed precision quantization technology. Apple has also conducted extensive evaluations of AFM's performance, including both human and automated evaluations, and the results show excellent performance in multiple benchmarks. Although developers can test these features, the official release may be delayed.

OpenAI launches GPT-4o long output mode, supporting up to 64k token output

OpenAI recently began testing the long output mode of its language model GPT-4o. This mode can support output of up to 64k tokens and is currently in the Alpha testing phase. Test users can access this new feature through the "gpt-4o-64k-output-alpha" API endpoint. Due to the higher cost of inference in long output mode, OpenAI has adjusted the cost for users using this model to $6 per million input tokens, while the cost per million output tokens has been increased to $18 to match the higher cost.

Meta releases SAM 2 to support real-time video tracking; AI Studio tool to create personalized AI chatbots

Meta announces Segment Anything Model 2 (SAM 2), a unified model for real-time, predictable object segmentation in both images and videos. SAM 2 is able to segment any object in any video or image, including previously unseen objects and visual domains, supporting a variety of use cases without the need for custom adaptations. To support the training of SAM 2, Meta has released a large annotated database of approximately 51,000 real-world videos and over 600,000 masklets, which is larger than existing video segmentation datasets. SAM 2 will be open source and free to use, and hosted on platforms such as Amazon SageMaker.

Meta also launched a new tool, AI Studio. This is a platform designed to allow users to create, share and customize AI characters without any technical skills. AI Studio is based on Meta's own Llama 3.1 model, allowing users to create freely and use the created AI characters on various social media platforms such as Instagram, Messenger, WhatsApp and web chat. Users can customize the name, personality, tone, avatar and slogan of the AI ​​character.

Zhiyoumai Technology establishes a partnership with large model company MiniMax

Worth Buying Technology and the large model company MiniMax officially announced a partnership to bring consumers a new shopping experience. It is reported that MiniMax's top product Conch AI has been connected to the "Consumer Large Model Enhanced Toolset" provided by Worth Buying, providing users with a more convenient way to compare prices and purchase. Through Conch AI's unique price comparison strategy, users can easily use the floating ball function to quickly compare the prices of different products and jump directly to the purchase page. This cooperation provides consumers with more choices and greatly reduces the cumbersome steps in the shopping process.

Visual China and Aishi Technology have reached a strategic cooperation to promote the development and application of AI video big models

Visual China and Aishi Technology recently announced a strategic partnership to jointly promote the development and application of AI video big models in terms of data cooperation and product collaboration. Aishi Technology was founded in April 2023 and focuses on AI video generation big models and their applications. Its product PixVerse AI video generation platform supports users to quickly generate video content through text and pictures. Visual China has a large number of professional-grade picture and text pairs and video music materials, as well as complete structured metadata and industry knowledge graphs. The cooperation will promote the continuous upgrading of video big models and improve the efficiency and quality of AI visual content generation.

NVIDIA and Hugging Face launch efficient inference service, increasing AI model token processing efficiency fivefold

The open source platform Hugging Face and NVIDIA have launched Inference-as-a-Service, which will be driven by NVIDIA's NIM technology. The launch of the new service will allow developers to prototype more quickly, use the open source AI models provided on the Hugging Face Hub, and deploy them efficiently.

Specifically, when accessed as NIM, a 7-billion-parameter Llama3 model can be processed five times faster than when deployed on a standard NVIDIA H100 Tensor Core GPU system, a huge improvement. In addition, the new service also supports "Train on DGX Cloud", which is currently available on Hugging Face.

JD.com launches eight products including Yanxi Intelligent Platform and JoyCoder, an intelligent programming assistant

JD Cloud has launched a series of large model application products. JD Cloud's enterprise large model service supports one-stop creation of enterprise-specific large models, while maintaining the injection of vertical field knowledge and the general capabilities of the model. Yanxi AI Development Computing Platform 2.0 provides a variety of open source models and data sets, as well as large model development services. In addition, JD Cloud has also launched a large model security and trust platform, covering more than 200 adversarial attack methods and covering all 31 types of risks required by regulatory compliance. JD Cloud Yunjian AI Computing Cloud supports unified scheduling of distributed computing power in multiple regions and provides cost-effective computing power supply. The new generation of distributed storage Yunhai 3.0 supports AI large models with hundreds of billions of parameters. JD Cloud Yanxi Intelligent Agent Platform has connected to dozens of large models, with more than 3,300 active agents within JD, and provides solution templates for the rapid implementation of industry applications.

todayProduct News

Product HuntHot list, AI video editing platform Topview.ai

Topview.ai is a free online AI video editing tool that can quickly convert user-provided links or media assets into attractive video ads. The platform analyzes more than 5 million popular videos on YouTube and TikTok, uses GPT-4o to generate scripts and storyboards, and uses AI avatars to enhance video effects. The AI ​​features provided by Topview.ai include script generation, clip selection and editing, voice narration, avatar creation, automatic subtitle generation, and multi-language support. These features can create attractive marketing videos for a variety of scenarios such as e-commerce marketing, product introductions, and app promotions. Users do not need professional video editing skills, and can use Topview.ai to improve video production efficiency and significantly reduce costs.

https://www.topview.ai/?ref=producthunt

Star Product News

1. Vidu video models are now available worldwide! Register now and use it without waiting in line, with a focus on supporting anime style

Vidu comes from Tsinghua-affiliated multimodal big model company Shengshu Technology. It fully opens up the text/image generation and video generation functions, and the generation time can be freely selected as 4s/8s, with a resolution of up to 1080P and a fast generation speed. The official claim is that it "achieves the fastest measured inference speed in the industry", and it only takes 30s to generate a 4s video clip.

Users can register to get 80 points per month, which can be used to generate videos for free. The excess requires a paid subscription. Vidu's highlight is that it supports character consistency and anime style, ensuring the consistency of the character's image and expression in different scenes, and can generate anime-style video clips. In addition, Vidu has significantly improved its reasoning speed, which is faster than the fastest level in the industry, and pursues high-quality video effects and special effects generation, as well as good processing capabilities for complex scenes.

https://www.vidu.studio/

2. Runway Gen-3 Alpha launches the image-to-video function

Runway released Gen-3 Alpha Image to Video. This update allows users to use any image as the first frame of a video generation, either on its own or by adding a question to the image to facilitate subsequent video generation. Image to Video is a major update that greatly improves artistic control and consistency of generation.

https://runwayml.com/ai-tools/gen-3-alpha/

Special attention

Future Pier: Huang Renxun and Zuckerberg discuss the future of AI and open source ecology

At the SIGGRAPH 2024 conference, NVIDIA founder and CEO Jensen Huang had an in-depth conversation with Meta founder and CEO Mark Zuckerberg.

Huang Renxun first asked Zuckerberg about the application of generative AI in advertising and business operations, and how to use these advances to introduce new features. Zuckerberg responded that Meta has worked in the field of SIGGRAPH for 8 years and completed research on hand tracking, codec avatars, and display systems, which is expected to promote the advancement of the metaverse and AI.

The two discussed the progress of recommendation systems and generative AI. Huang Renxun emphasized the importance of recommendation systems as one of the largest computing systems in the world, while Zuckerberg mentioned that Meta is working hard to create a more general recommendation model to improve system efficiency and content personalization. He also shared some of Meta's vision in the field of AI, including empowering everyone who uses Meta products to create their own AI agents, and the concept of AI Studio, which allows users to create and customize their own versions of AI.

Zuckerberg further explained Meta's open source philosophy, including open source PyTorch and LLaMA, as well as his views on open computing platforms. He believes that open AI models will restore the entire industry's ecosystem to a more open and diverse level. Huang Renxun mentioned NVIDIA's HGX system and the concept of AI foundry, and how to help companies build their own AI services.

The two also discussed the need for smaller AI models and how this need can be met through distillation and fine-tuning of large models. Finally, they explored the future development of smart glasses and mixed reality technology and how these new computing devices can be combined with AI to provide consumers and enterprises with new ways of interacting and working.

https://www.nvidia.com/en-us/events/siggraph/

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.

AI image generation platform LiblibAI raises hundreds of millions of yuan in financing; Meta reveals that training Llama 3 "strikes" every 3 hours | AI Intelligence Bureau

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