Today's Financing Express
AI Agent Platform DevRev Receives $100 Million in Financing
DevRev has completed a $100 million Series A financing round, with the company's valuation reaching $1.15 billion. This round of financing was supported by several well-known investment institutions, including Khosla Ventures and Mayfield Fund.
DevRev's core product, the AgentOS platform, can easily extract data from traditional systems, help users develop lightweight AI agents, and achieve real-time data connections through powerful knowledge graphs. The AgentOS platform has powerful search and analysis capabilities, and can be seamlessly integrated with existing solutions such as Zendesk and Salesforce. In terms of code development, AI Agent can quickly analyze large code bases and user feedback, help developers quickly identify problems, accelerate the development process, and provide personalized code improvement suggestions. Currently, DevRev has more than 1,000 corporate customers, including several well-known SaaS unicorns and one of the world's top five consumer banks.
AI enterprise search platform Glean is valued at $4.5 billion, DST Global may invest
Glean is an enterprise AI search tool with an experience very similar to ChatGPT, but limited by the boundaries of enterprise data and content. Users can index dozens of applications, understand context, language, behavior, and employee relationships, and find personalized answers to questions.
Glean is currently working with DST Global on a new round of $250 million in financing, with its valuation tripling to $4.5 billion compared to six months ago. As of February this year, Glean has more than 200 corporate customers, including Instacart, Pinterest, Reddit, Duolingo, Grammarly, Webflow, Confluence and Sony.
AI defense company Anduril raises $1.5 billion in funding, latest valuation at $14 billion
Anduril is a US defense technology company focused on developing advanced AI and automation technologies to enhance national security and defense capabilities. The investment was co-led by Founders Fund and Sands Capital. Fidelity Investments, Baillie Gifford and some other institutional investors also joined. Anduril is now valued at $14 billion, up from $8.5 billion after its last round of financing in 2022.
Tecorigin completes A2 round of financing worth hundreds of millions of yuan, with Golden Ant Investment and Frost Leaf Ventures participating in the investment
Tecorigin has completed a multi-million-yuan A2 round of financing, with participation from Zhejiang Province’s state-owned financial holding platform Jinma Investment and Shuangye Venture Capital.
The Taichu team builds high-performance, high-efficiency, and highly reliable intelligent computing systems for government and enterprise users, deeply customizes HPC+AI key field solutions, and covers the entire process of intelligent computing center construction, including software and hardware research and development, computing system design and integration.
SoundHound acquires Amelia for $80 million to expand its conversational voice AI solutions
SoundHound AI Inc. acquires Amelia, an enterprise conversational AI agency, to expand its business conversational and voice AI solutions. SoundHound AI agreed to pay $80 million in cash and equity for Amelia. According to Pitchbook data, Amelia has raised at least $189 million since its inception, including a $175 million round led by BuildGroup and Monroe Capital in March 2023.
SoundHound develops voice AI technology that understands natural speech and allows people to interact with technology such as vehicles, services, devices and applications. It also provides companies with a voice-enabled digital assistant built with generative AI and has launched a music recognition mobile app.
Robot laser vision system developer Yinglai Technology completes 20 million yuan financing
Yinglai Technology is mainly engaged in the research and development and production of laser vision systems. At the same time, it cooperates with Tangshan Yinglai Machine Vision Engineering Technology Application Center to provide users with solutions for laser vision sensors for industrial robots. Recently, Yinglai Technology received RMB 20 million in round B equity financing from Wujiang Dongfang State-owned Assets.
AI app Feeling Great raises $8 million in seed funding
The Feeling Great app was developed by psychiatrist and co-founder David Burns to translate his 40 years of psychotherapy experience into artificial intelligence, providing interactive mental health sessions and chatbot services to help those who may not have access to professionals.
Founded in 2022, the company has raised $8 million in seed funding co-led by Learn Capital and TitletownTech, with other investors including Lux Ventures, Wavemaker Three-Sixty Health, Pacific Health Ventures, and Treble Capital.
AI therapy startup clare&me raises €3.7 million
Clare&Me is a Berlin-based AI mental health tech startup that has raised €3.7 million in funding. Their chatbot is designed to be used in clinical settings as a conversational model for patients to discuss their mental health conditions.
The round was led by Munich-based HealthTech investor YZR, with participation from new investors including Kodori Ventures and Material Ventures, as well as participation from existing angel investors and Antler. clare&me has also received grant funding from IBB (Berlin Development Bank), European funds and the Federal Ministry of Education and Research to continue technology research and develop the world’s first CLLM (Clinical Large Language Model), which will be used by psychotherapists for training.
375ai raises $5 million in seed funding for decentralized edge data intelligence network
375ai is committed to building a large-scale decentralized edge data intelligence network, focusing on innovative AI-driven data transmission, aiming to change the way edge data is collected, analyzed and monetized. The company has officially launched its first product, 375edge, an advanced edge AI node designed to efficiently and accurately collect and analyze multimodal data in real time. This round of financing was led by 6th Man Ventures, Factor, Arca, EV3, Primal Capital and Auros.
AI video creation platform Reforged Labs receives $3.9 million in funding
Reforged Labs has raised $3.9 million to use AI to power gaming ads. Reforged Labs is an AI video creation platform that enables businesses to quickly create video ads with customization options without any video production experience.
AI business agency Bardeen receives strategic investment from Dropbox and Hubspot
Bardeen's platform uses a natural language interface to automate repetitive knowledge work. The agent platform is a browser extension that is context-aware so that the agent can perform the "planning" step after receiving instructions from the user.
The company has raised $3 million in a new round of funding, bringing its total funding to $22 million. Dropbox and HubSpot have become strategic investors in the startup through their venture capital arms (Dropbox Ventures and HubSpot Ventures).
AI liquidity distribution protocol Quantlytica completes seed round financing
This round of financing was jointly invested by Connectico Capital, DWF Labs, Eureka Partners, Polygon Labs, Web3port Foundation and ZBS Capital. Quantlytica is an AI liquidity allocation protocol dedicated to simplifying the search and construction of the best funding strategy portfolio, aiming to solve the Web3 liquidity allocation problem.
Sequoia Capital USA may invest in Harmonic, an AI inference company founded by Robinhood CEO
Harmonic is a recent entrant into the AI space, developing AI technology that can reason through problems. The company was co-founded by Robinhood CEO Vlad Tenev and self-driving car entrepreneur Tudor Achim.
Harmonic is reportedly in discussions with investors including Sequoia Capital and Index Ventures for funding. According to a person familiar with the matter, the round of financing could total $5 million, with a pre-money valuation of $250 million. Harmonic has previously received early investments from Quora founder Adam D'Angelo, DST Global's Yuri Milner, and Ribbit Capital founder Micky Malka.
Hugging Face acquires XetHub
Hugging Face announced the acquisition of XetHub, founded by Yucheng Low, Ajit Banerjee, and Rajat Arya, who previously worked at Apple to build and scale its internal machine learning infrastructure.
XetHub's mission is to provide software engineering best practices for AI development, and they have developed technology that enables Git to scale to terabyte-scale repositories and enables teams to collaborate more efficiently on large datasets and models. The XetHub team will help Hugging Face unlock growth over the next five years by using its own version of LFS as the storage backend for the Hub repository.
Nvidia-backed Recursion acquires Exscientia, a provider of AI technology services for drug discovery and development
Exscientia's main business is to use the developed artificial intelligence platform to guide the development of automated drugs, and use big data and machine learning methods to automatically design millions of small molecule compounds related to specific targets based on existing drug development data, and evaluate and screen the compounds based on other conditions such as efficacy, selectivity, ADME, etc., and then conduct experimental tests on the screened compounds and feed them back to the AI system for screening, aiming to shorten the new drug development process. Pharmaceuticals has agreed to acquire smaller competitor Exscientia in an all-stock transaction for $688 million.
Box acquires AI document processor Alphamoon
Cloud content management company Box Inc. has acquired Alphamoon’s artificial intelligence document processing technology. Box did not disclose the terms of the acquisition, but said the company will expand the capabilities of the Box Intelligent Content Cloud by integrating Alphamoon’s AI technology with Box’s existing AI capabilities.
(Welcome to add WeChat
AIyanxishe2
to learn more about AIGC and financing, and chat about new AI products with like-minded friends)
Microsoft Copilot AI is said to be manipulated by hackers to steal corporate secrets
Michael Bargury, co-founder and CTO of security company Zenity, disclosed at the Black Hat security conference that Microsoft's Copilot AI built into the Windows system has security risks. These vulnerabilities can be exploited to obtain sensitive user information, such as contact data, and can be used to forge phishing emails to trick corporate employees into performing unsafe operations, such as modifying bank transfer information.
The researchers demonstrated how hackers can use Copilot to leak data and conduct phishing attacks. Copilot Studio allows companies to customize chatbots, but this requires AI to access corporate data, which creates security risks. Attackers can bypass Copilot's protection measures by injecting malicious data. Bargury pointed out that when AI is given access to data, this data becomes a target for attack.
AI software developer Anaconda sues Intel for infringement: license expired but not renewed but still used
AI software developer Anaconda accused Intel of continuing to use Anaconda software to develop an artificial intelligence platform after its license expired and did not respond to Anaconda's renewal proposal. Anaconda believes that Intel deliberately used its technological innovations to improve its own products and gain an advantage in the artificial intelligence market. Anaconda's software is free for individuals and small businesses to use. The lawsuit also mentioned that Intel illegally integrated Anaconda software into its own AI toolkit and distributed it to other developers without permission. Anaconda asked Intel to stop abusing its software and pay compensation, but lawyers for Intel and Anaconda declined to comment on the case.
Alibaba and the National University of Singapore jointly launched CodexGraph to allow AI to manage large code bases
Researchers from the National University of Singapore, Alibaba, and Xi'an Jiaotong University jointly developed CodexGraph, which improves the processing capabilities of large-scale, multi-file code bases by building a code graph database and connecting language models with code bases. The system's workflow includes static analysis of code symbols and relationships, code structure-aware queries, "Write then translate" strategies, and iterative pipelines.
Evaluations on three academic benchmarks show that CodexGraph exhibits excellent and stable performance in cross-file code completion, code generation, and automated GitHub issue resolution. In addition, CodexGraph has been put into practical use in five real-world code application scenarios, such as code chat assistants and code debugging tools, improving the efficiency and quality of code development. The researchers expect CodexGraph to support more programming languages in the future and further optimize workflows and index building efficiency.
Product Hunt
Hot List, AICamp
AICamp is a solution that provides enterprises with a personalized artificial intelligence platform that enables team members to leverage internal knowledge, accelerate workflows, and build customized AI assistants.
AICamp provides multiple features, including: Chat function for communicating with various models or assistants; Assistant function for building custom assistants to communicate with internal resources; upcoming Workflow and Integrations functions, as well as existing Teamspaces, Extensions, custom API integration, model selection, internal knowledge connection, personal prompt library management, effective collaboration, and role permission management. In addition, AICamp also provides detailed AI usage analysis, suitable for organizations of all sizes.
???? https://aicamp.so/?ref=producthunt
Github Trending Hot List,
Deep-Live-Cam
A photo, live broadcast of real-time face-changing celebrities
The Deep-Live-Cam project makes face-swapping possible during live broadcasts. The project is open source on GitHub, supports multiple hardware accelerations, and has received a lot of attention in a short period of time. Blogger Matthew Berman conducted tests and verified its effects under different lighting conditions, and even better under low-light conditions. Deep-Live-Cam is easy to operate and has a clear interface, making it easy for ordinary people to use. Although Deep Face Live also provides similar functions, Deep-Live-Cam performs more naturally in generating face-swapping effects in a single photo.
????
https://github.com/hacksider/Deep-Live-Cam
1. Qwen2-Math, Alibaba’s special language model for mathematics
Alibaba's Qwen team released a new series of mathematical language models, Qwen2-Math, based on Qwen2. The flagship model, Qwen2-Math-72B-Instruct, performs well in math-related downstream tasks, surpassing several advanced models including GPT-4o, Claude 3.5-Sonnet, Gemini-1.5-Pro, and Llama-3.1-405B.
???? https://github.com/QwenLM/Qwen2-Math
2. Apple open-sources new image generation model ml-mdm
Apple introduced an image generation model and training method called Matryoshka Diffusion Models (MDM). The model only uses the CC12M dataset containing 12 million images and demonstrates strong zero-sample generalization capabilities. The name MDM comes from Russian nesting dolls, which nest small structures within large structures, similar to each doll being independent and complete. MDM can process images at different resolutions simultaneously, achieving seamless generation from low-definition sketches to high-definition details.
????
https://github.com/apple/ml-mdm
Nick Frosst, Cohere co-founder: AI is unlikely to reach general artificial intelligence
Nick Frosst, co-founder of Cohere, believes that although artificial intelligence companies are rapidly attracting investment and obtaining high valuations, this does not mean that the industry is in a bubble. Frosst pointed out that Cohere has helped its enterprise customers realize new functions and process automation that were previously impossible by building custom AI models for them, which proves the practical value of AI technology.
He is skeptical about the possibility of AI reaching general artificial intelligence, believing that AI will not become a "digital god" but a powerful and useful tool. Cohere's business model is based on Aidan Gomez's research, including the idea that a large language model can provide more value. Frosst also mentioned that companies should focus on the practical application of AI technology rather than expecting a single model to solve all problems. He emphasized that Cohere is cautious about the practicality of AI technology and believes that AI will not lead to the extinction of mankind.
???? https://techcrunch.com/2024/08/08/cohere-co-founder-nick-frosst-thinks-everyone-needs-to-be-more-realistic-on-what-ai-can- and-cannot-do/
Stay tuned for the latest updates tomorrow
!
AI Intelligence Bureau 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:
AIyanxishe2
and note the industry position.