HomeArchivePricing

GenAI Daily Report 2024-03-13

News Express

  • Cognition AI Introduces Devin, an AI Software Engineer

  • Covariant Launches RFM-1, Aiming to Revolutionize Robotics

  • OpenAI upcoming GPT-4.5 Turbo with Enhanced Features

  • Cohere Launches Command-R: A 35B Model Optimized for RAG

  • DoorDash Introduces 'SafeChat+' to Curb Verbal Abuse

  • Reddit's IPO Price Justified by AI Revenue Potential

  • Locus Robotics Leverages AI in Warehouse Automation

  • Deepgram Launches Aura: Real-Time Text-to-Speech API

  • Pienso Raises $10M for No-Code AI Model Training Tools

  • Truffle-1 Launches: Affordable Inference Engine for LLMs

  • Meta Launches Two New 24K GPU Clusters for AI

  • Walmart Enhances Retail with Element AI Platform

Industry

Google, Meta, NYT, IBM, Apple Make AI Headlines

Key updates on AI from Google, Meta, NYT vs. OpenAI, IBM layoffs, and Apple's ML project showcase the technology's evolving landscape.

Detail:

  • Google limits its Gemini AI's responses on election-related queries globally to prevent misinformation, particularly with the Lok Sabha elections in focus.
  • Meta announces the launch of two new 24k GPU clusters featuring Nvidia H100 GPUs to bolster AI research and training capabilities, aiming for significant expansion by end of 2024.
  • The New York Times disputes OpenAI's 'hacking' claims in a copyright lawsuit, asserting it used initial sentences of articles to prompt ChatGPT, not hacking.
  • IBM plans to reduce its workforce by 80% in certain departments, including HR, as part of its pivot towards artificial intelligence technologies.
  • Apple introduces MLXServer with Swift, simplifying API creation for ML models, marking a significant step in making machine learning more accessible to developers.

URL: https://twitter.com/chiefaioffice/status/1767709702333346255 From: twitter chiefaioffice

OpenAI upcoming GPT-4.5 Turbo with Enhanced Features

OpenAI's new GPT-4.5 Turbo model surpasses its predecessor in speed, accuracy, and scalability.

Detail:

  • OpenAI has announced the upcoming release of GPT-4.5 Turbo, a model that improves upon GPT-4 Turbo.
  • GPT-4.5 Turbo is noted for its enhanced speed, accuracy, and scalability.
  • The model can generate natural language and code with a 256k context window.
  • An anticipated release or significant update is hinted for June 2024.

URL: https://twitter.com/chiefaioffice/status/1767616019671986356 From: twitter chiefaioffice

Covariant Launches RFM-1, Aiming to Revolutionize Robotics

Covariant introduces RFM-1, a platform designed to endow robots with human-like reasoning and adaptability across various industries.

Detail:

  • RFM-1 is a large language model for robotics, designed to provide robots with human-like reasoning abilities.
  • Developed by Covariant, RFM-1 utilizes extensive real-world data to support a wide range of applications from manufacturing to domestic use.
  • The platform represents a departure from single-purpose robots, enabling more flexible and adaptable robotic solutions.
  • RFM-1's introduction aligns with the growing interest in general-purpose robotic systems and aims to complement advancements in robotic hardware.
  • Covariant's approach focuses on natural interaction with robots, potentially simplifying the programming process and enhancing operational efficiency.

URL: https://twitter.com/DrJimFan/status/1767589636778701149 From: twitter DrJimFan

Cognition AI Introduces Devin, an AI Software Engineer

Cognition AI has unveiled Devin, an AI that excels in software engineering, marking a significant advancement in autonomous coding technology.

Detail:

  • Devin is heralded as the first AI software engineer by Cognition AI.
  • It has outperformed on the SWE-Bench coding benchmark and passed engineering interviews from top AI firms.
  • Devin can autonomously complete coding tasks, including writing full applications and resolving issues.
  • This breakthrough suggests a future where developers could significantly enhance their productivity by working alongside AI engineers.
  • Jensen Huang's statement about the diminishing necessity for humans to learn coding and Sam Altman's discussion on the potential for very small teams to achieve monumental success are highlighted as relevant to this development.

URL: https://twitter.com/saranormous/status/1767564859603030352 From: twitter saranormous

Cohere Launches Command-R: A 35B Model Optimized for RAG

Cohere unveils Command-R, a 35B model with special optimizations for RAG, outperforming GPT-3.5 in RAG capabilities.

Detail:

  • Command-R is a 35B model, open-sourced by Cohere, designed with a 128K context specifically optimized for RAG, surpassing GPT-3.5 in RAG capabilities.
  • The model demonstrates higher preference in enterprise RAG use cases and accuracy in KILT end-to-end RAG evaluations compared to other models.
  • Command-R is scalable, focuses on enabling enterprises to deploy AI at production scale with high accuracy, low latency, extended context, and supports 10 key languages.

URL: https://twitter.com/Gorden_Sun/status/1767566901373473007 From: twitter Gorden_Sun

Key Developments and Innovations in AI Announced

Recent announcements highlight significant advancements and new features across various AI domains, emphasizing the sector's rapid evolution.

Detail:

  • Cohere introduces Command-R, a scalable generative AI model supporting 10 languages for enterprise deployment.
  • Sam Altman cleared of misconduct, rejoining OpenAI's board along with new members from Gates Foundation, Sony, and Instacart.
  • Elon Musk announces xAI will open source Grok, marking a significant moment for open-source projects in AI.
  • Apple's Siri is set for a major upgrade to become a formidable competitor in the AI assistant race against ChatGPT.
  • South Korea introduces a $1,800 AI-powered companion doll to address loneliness among the aging population.
  • Covariant unveils RFM-1, an AI platform enabling ChatGPT-like language reasoning for physical robots.
  • Midjourney releases a new feature for character consistency in outputs, showing the ongoing innovation in AI-generated content.

URL: https://twitter.com/rowancheung/status/1767410913622003916 From: twitter rowancheung

DoorDash Introduces 'SafeChat+' to Curb Verbal Abuse

DoorDash launches 'SafeChat+', an AI tool to automatically detect and manage verbal abuse in in-app conversations, aiming to enhance safety.

Detail:

  • 'SafeChat+' utilizes AI to identify offensive language in real-time, covering multiple languages.
  • Options for reporting incidents include contacting support for customers and order cancellation for delivery persons without affecting their ratings.
  • DoorDash will issue warnings to users for inappropriate conduct and has a team to investigate all AI-identified incidents.
  • The tool is an advancement from SafeChat, offering more sophisticated technology to recognize subtle threats.
  • DoorDash's initiative aims to decrease verbal abuse incidents on its platform, contributing to a safer environment for both consumers and Dashers.

URL: https://techcrunch.com/2024/03/12/doordashs-new-ai-powered-safechat-tool-automatically-detects-verbal-abuse/ From: techcrunch Lauren Forristal

Agility Robotics Showcases Digit's Capabilities at Modex 2024

At Modex 2024, Agility Robotics highlighted Digit's advancements and its integration into warehouse and manufacturing workflows, underlining the company's leadership and innovation.

Detail:

  • Agility Robotics stood out at Modex 2024 with Digit, showcasing its advancements in handling tasks in warehouse and automotive manufacturing settings.
  • The company demonstrated Digit's capabilities in lineside replenishment and tote retrieval, highlighting ongoing collaborations with automotive industry players.
  • Agility has shifted its focus towards addressing immediate warehouse staffing issues, with last-mile delivery still being a consideration for the future.
  • Leadership changes, including the appointment of Peggy Johnson as CEO and other key C-suite roles, underscore Agility's push towards commercialization and gender diversity in leadership.
  • Agility introduced Agility Arc, a software platform for managing Digit fleets, emphasizing integration with existing warehouse management systems for seamless operations.
  • Future developments for Digit include the potential for swappable end effectors, enhancing its adaptability to different tasks without the need for complex hand mechanisms.

URL: https://techcrunch.com/2024/03/11/the-loneliness-of-the-robotic-humanoid/ From: techcrunch Brian Heater,@bheater

Reddit's IPO Price Justified by AI Revenue Potential

Reddit's planned IPO pricing seems high but is backed by its significant AI revenue potential, justifying its valuation.

Detail:

  • Reddit's S-1 filing suggests an IPO price of $31 to $34 per share, aiming for a valuation around $5.4 billion.
  • The company's valuation could exceed $6 billion when considering options held by employees and others.
  • With $804 million in 2023 revenue, Reddit's valuation represents 6.9x to 8x its revenue.
  • Reddit's $203 million contracts with AI companies for data access highlight its AI revenue potential.
  • OpenAI CEO Sam Altman, owning an 8.7% stake in Reddit, underscores the company's significant role in AI data training.
  • Despite being unprofitable, Reddit's AI data monetization strategy is a key factor in its IPO pricing.

URL: https://techcrunch.com/2024/03/11/reddit-ipo-share-price-plans/ From: techcrunch Alex Wilhelm

Elon Musk Sues OpenAI Over Management Disputes

Elon Musk's lawsuit against OpenAI sparks a broad discussion among tech investors about the company's direction and the future of AI technology.

Detail:

  • Elon Musk has filed a lawsuit against OpenAI, citing a departure from its founding principles.
  • The lawsuit has ignited a debate among tech investors, with some viewing it as unjustified.
  • Discussions also explore the impact on open source AI development and Meta's role in the AI industry.
  • This legal battle underscores the intense competition and regulatory scrutiny surrounding AI technology.

URL: https://techcrunch.com/2024/03/11/watch-elon-musk-sam-atlman-and-the-rest-of-the-billionaires-are-fighting-over-the-future-of-ai/ From: techcrunch Alex Wilhelm

Locus Robotics Leverages AI in Warehouse Automation

Locus Robotics emphasizes its software's role in setting it apart in the warehouse automation sector, especially with its new AI-driven platform.

Detail:

  • Locus Robotics, known for its autonomous mobile robots, is fundamentally a software company, according to CEO Rick Faulk.
  • The company's software, particularly its fleet management software and the newly announced LocusHub Engine, is crucial for orchestrating robotic systems in warehouses.
  • LocusHub Engine utilizes AI to process data collected by robots, offering predictive analytics for warehouse management, such as staff distribution and identifying bottlenecks.
  • Locus' foundation was a response to Amazon's acquisition of Kiva Systems, aiming to fill the service gap left for external clients.
  • Despite the market's fluctuating interest and a recent minor layoff, Locus remains a leader in warehouse automation, focusing on existing client needs and steady growth.

URL: https://techcrunch.com/2024/03/12/locus-robotics-success-is-a-tale-of-focusing-on-what-works/ From: techcrunch Brian Heater,@bheater

Deepgram Launches Aura: Real-Time Text-to-Speech API

Deepgram introduces Aura, a new API for developing conversational AI agents with realistic voice models and low latency at competitive prices.

Detail:

  • Deepgram, known for voice recognition, announces Aura, a text-to-speech API designed for real-time, conversational AI.
  • Aura features highly realistic voice models with low-latency, aiming to replace customer service agents in various sectors.
  • Scott Stephenson, Deepgram's CEO, emphasizes Aura's combination of accuracy, speed, and affordability.
  • Aura's pricing is competitive at $0.015 per 1,000 characters, slightly cheaper than Google's WaveNet and Amazon's Polly.
  • The product offers around a dozen voice models, all trained in-house using a dataset created with voice actors.

URL: https://techcrunch.com/2024/03/12/deepgrams-aura-gives-ai-agents-a-voice/ From: techcrunch Frederic Lardinois,@fredericl

Pienso Raises $10M for No-Code AI Model Training Tools

Pienso, a startup offering no-code AI model training platforms, raises $10 million in Series A funding to facilitate AI deployment for non-technical users.

Detail:

  • Pienso, founded by MIT alumni, simplifies AI model training with a no-code platform, targeting non-technical talent such as researchers and marketers.
  • The platform allows users to annotate data for training AI models without coding, which can be deployed in the cloud or on-premises.
  • It addresses common AI deployment challenges like data management and security, enabling domain experts to train models with their own data.
  • Pienso's business model offers flexible licensing based on the number of deployed AI models, encouraging experimentation before full deployment.
  • The $10 million Series A funding led by Latimer Ventures will be used for scaling the company's operations and developing new features.

URL: https://techcrunch.com/2024/03/12/pienso-is-building-no-code-tools-for-training-ai-models/ From: techcrunch Kyle Wiggers

Truffle-1 Launches: Affordable Inference Engine for LLMs

Truffle-1 is announced, a $1299 inference engine capable of running large language models at 60 watts.

Detail:

  • Truffle-1 can operate at 20 tokens/s using Mixtral 8x7B, scaling up to 100B models.
  • It's designed for running open-source models efficiently with a power consumption of only 60 watts.
  • Available for preorder at $1299.

URL: https://twitter.com/Gorden_Sun/status/1767674337958781205 From: twitter Gorden_Sun

Meta Launches Two New 24K GPU Clusters for AI

Meta has unveiled two new 24K GPU clusters aimed at advancing generative AI development.

Detail:

  • Meta announces the introduction of two 24K GPU clusters.
  • The clusters are designed to enhance generative AI development.
  • This development was reported by SiliconANGLE.

URL: https://twitter.com/robhof/status/1767717239841751084 From: twitter robhof

Snowflake and Nvidia Discuss AI, Data, and Cloud Intersection

Snowflake and Nvidia executives share insights on the convergence of AI, data analytics, and cloud computing.

Detail:

  • Executives from Snowflake and Nvidia discuss the importance of integrating AI, data analytics, and cloud computing.
  • The conversation highlights how this intersection is essential for the future of technology and business strategies.
  • Insights include the impact on innovation, efficiency, and scalability in the tech industry.

URL: https://twitter.com/robhof/status/1767664310472655007 From: twitter robhof

Walmart Enhances Retail with Element AI Platform

Walmart has developed the Element AI platform to revolutionize retail through data harnessing.

Detail:

  • Walmart introduces Element AI to optimize retail operations.
  • The platform aims at enhancing customer experience and operational efficiency.
  • Element AI leverages advanced data analytics for strategic decision-making.

URL: https://twitter.com/robhof/status/1767633444614672842 From: twitter robhof

Managing Impacts of US Executive Order on AI for Leaders

The article provides strategies for executive leaders to navigate the challenges posed by the US executive order on AI.

Detail:

  • Highlights the significance of the US executive order on AI for businesses.
  • Outlines practical strategies for executive leaders to mitigate risks and leverage opportunities.
  • Emphasizes the importance of staying informed and agile in response to regulatory changes.

URL: https://twitter.com/robhof/status/1767562221226524961 From: twitter robhof

Akamai Teams Up with Neural Magic for AI Workloads on Edge Servers

Akamai and Neural Magic collaborate to enhance AI workload performance on edge CPU servers.

Detail:

  • The partnership aims to accelerate AI workloads on edge CPU servers.
  • This collaboration seeks to leverage the strengths of both companies in improving AI performance.
  • The initiative underscores the growing importance of edge computing in AI applications.

URL: https://twitter.com/robhof/status/1767545778120364497 From: twitter robhof

Automakers May Secretly Sell Your Driving Data to Insurers

A report suggests that automakers could be covertly selling driving data to insurance companies, raising privacy concerns.

Detail:

  • A new report highlights potential privacy issues with automakers possibly selling driving data to insurance companies without explicit consent from car owners.
  • The article emphasizes the implications this practice could have on privacy and the lack of transparency in how driving data is used.
  • Concerns are raised regarding how this data selling could affect insurance rates and consumer trust in automakers.

URL: https://twitter.com/robhof/status/1767383935712162283 From: twitter robhof

Midjourney Launches New Character Reference Feature

Midjourney introduces a Character Reference feature, enhancing user control over character consistency in generated images.

Detail:

  • The Character Reference feature allows users to generate images with consistent characters by using a reference image URL with the --cref command.
  • Users can adjust the 'strength' of the reference from 100 to 0 using the --cw command, affecting how closely generated characters match the reference in terms of face, hair, and clothes.
  • This feature is similar to the previously available Style Reference tool but focuses specifically on matching character traits.
  • Advanced options include blending traits from multiple characters by using multiple URLs, similar to using multiple image or style prompts.

URL: https://twitter.com/FinanceYF5/status/1767444865972044114 From: twitter FinanceYF5

OpenAI VP Dodges Compensation Question for Artists at SXSW

OpenAI's VP, Peter Deng, avoided directly answering whether artists should be compensated for their work used in AI training at SXSW.

Detail:

  • Peter Deng, OpenAI's VP, refrained from directly answering if artists should be compensated for their work used to train AI, like ChatGPT, during SXSW.
  • Some audience members agreed artists should be compensated, a sentiment acknowledged by Deng.
  • OpenAI's stance is that using publicly available internet materials for AI training falls under fair use, a view not universally accepted, especially by creators.
  • A class action lawsuit by artists against OpenAI and its rivals highlights the controversy over AI-generated works that replicate artists' styles without compensation.
  • OpenAI has some licensing agreements, e.g., with Shutterstock, and offers an opt-out for artists wanting to remove their work from training datasets, though the process is criticized as burdensome.
  • Deng expressed a belief in giving artists more agency in generative AI development but was uncertain about the specific implementations.

URL: https://techcrunch.com/2024/03/11/should-artists-be-paid-for-training-data-openai-vp-wouldnt-say/ From: techcrunch Kyle Wiggers

Launch of New Company Physical Intelligence

A group of individuals including @saranormous have started a company called Physical Intelligence, aiming to introduce general-purpose AI to the physical world.

Detail:

  • Physical Intelligence, a new company, was co-founded by @saranormous, @lachygroom, @hausman_k, @brian_ichter, @chelseabfinn, @svlevine, @QuanVng, and @SurajNair_1.
  • The company's mission is to bring general-purpose AI technologies into the physical world, bridging the gap between digital and physical realms.

URL: https://twitter.com/saranormous/status/1767715789203988920 From: twitter saranormous

'Become-Image': A New Replicate Model by Gorden Sun

Gorden Sun releases 'become-image', a Replicate model for blending faces and styles, with open-source code and workflow.

Detail:

  • Gorden Sun announced the release of 'become-image', a new model on Replicate.
  • 'Become-image' blends faces and styles to create unique images.
  • The project's code and ComfyUI workflow are open-sourced on Github.
  • Links to the model, code, and workflow are provided in the thread.

URL: https://twitter.com/Gorden_Sun/status/1767553759117291553 From: twitter Gorden_Sun

Midjourney's Character Consistency Outperforms SD

Midjourney's ability to maintain character consistency yields better results than SD, with a high success rate.

Detail:

  • Comparison highlights Midjourney's character consistency as superior to SD.
  • The effectiveness of Midjourney is attributed to its high probability of yielding satisfactory results despite an element of randomness.
  • Visual representations in provided images showcase the capability of Midjourney in creating characters with coherent aesthetics and themes.

URL: https://twitter.com/Gorden_Sun/status/1767581200934453483 From: twitter Gorden_Sun

Apple Softens App Store Rules in EU

Apple has made its App Store software distribution rules more lenient in the European Union.

Detail:

  • The change is specifically aimed at the European Union market.
  • This decision by Apple could affect how software is distributed and accessed via the App Store in EU countries.
  • The news was reported by SiliconANGLE, highlighting the potential implications for developers and consumers.

URL: https://twitter.com/robhof/status/1767693349014962211 From: twitter robhof

Evolution of Uber's AI-Driven Platform

An article details the development of Uber's AI platform, from initial concept to its current form.

Detail:

  • The article discusses the journey of Uber's AI platform.
  • It highlights the key stages from concept to the operational code.
  • The platform's impact on Uber's services and operations is examined.

URL: https://twitter.com/robhof/status/1767673880368615875 From: twitter robhof

MLOps and Generative AI Transform Organizational Strategies in 2024

In 2024, MLOps and generative AI are significantly transforming how organizations approach their AI strategies.

Detail:

  • The article highlights the growing importance of MLOps and generative AI in shaping future organizational AI strategies.
  • It suggests that these technologies are pivotal for the efficiency and innovation of AI deployments in 2024.
  • The discussion is based on insights shared during the Supercloud6 conference.

URL: https://twitter.com/robhof/status/1767665114206163246 From: twitter robhof

Zscaler Employs AI for Zero-Trust Security Enhancement

Zscaler utilizes AI technology to strengthen its zero-trust security framework, as reported by SiliconANGLE.

Detail:

  • Zscaler integrates artificial intelligence to advance its zero-trust security approach.
  • The enhancement aims at providing more robust and efficient cyber defense mechanisms.
  • This move reflects the evolving landscape of cybersecurity, where AI plays a crucial role.

URL: https://twitter.com/robhof/status/1767651339222552763 From: twitter robhof

'AI Everywhere' Theme Dominates Supercloud 6 Event

Supercloud 6 event analysis highlights the transformative impact of ubiquitous AI across technology sectors.

Detail:

  • Rob Hof reports on Supercloud 6, emphasizing the pervasive role of AI.
  • The analysis indicates AI's transformative effect on various technology domains.
  • Coverage by SiliconANGLE provides detailed insights from the event.

URL: https://twitter.com/robhof/status/1767630391945695693 From: twitter robhof

Investment

Wiz Inc.

Description: Wiz Inc. is an Israeli cloud security startup founded by Assaf Rappaport, Ami Luttwak, Yinon Costica, and Roy Reznik. Their software platform helps companies detect vulnerabilities and malicious activity in their cloud setup across network, identity, compute, application, and secrets for quicker insights and informed decision-making.
Amount: $800M, Round: new funding round
URL: https://siliconangle.com/2024/03/11/israeli-cloud-security-startup-wiz-reportedly-talks-raise-800m-10b-valuation/

Applied Intuition

Description: Applied Intuition accelerates the world's adoption of safe and intelligent machines through advanced driver-assistance systems and automated driving solutions. With $1 billion investment, the company has seen sustainable growth across automotive, trucking, defense, and agriculture industries.
Amount: $250M, Round: Series E
URL: https://www.appliedintuition.com/blog/series-e/?utm_source=ainews.bunnypixel.com

Applied Intuition

Description: Applied Intuition accelerates the world's adoption of safe and intelligent machines through vehicle software products and AI capabilities. With a $6 billion valuation and $1 billion in investment, the company serves global automotive, trucking, defense, construction, mining, and agriculture industries.
Amount: $250M, Round: Series E
URL: https://www.appliedintuition.com/blog/series-e

Applied Intuition

Description: Applied Intuition creates software that automakers and others use to develop autonomous vehicle solutions. The company works with 18 of the top 20 automakers, as well as autonomous vehicle startups and has a contract with the Army and Defense Innovation Unit.
Amount: $250M, Round: Series E
URL: https://techcrunch.com/2024/03/12/applied-intuition-series-e-raise-funding-andreessen-lux/

Bear Robotics

Description: Bear Robotics, founded in 2017, manufactures self-driving indoor robots to assist employees in various industries globally. Their flagship Servi robots are deployed in hospitality, healthcare, retail, and logistics sectors, easing daily tasks and enhancing efficiency.
Amount: $60M, Round: Series C
URL: https://www.businesswire.com/news/home/20240312368593/en/Bear-Robotics-Lands-60-Million-in-Series-C-Funding-from-Tech-Giant-LG-Electronics

Ocient

Description: Ocient, a Chicago-based startup founded in 2016, offers a Hyperscale Data Warehouse for efficient storage and analysis of data at scale. Their Compute Adjacent Storage Architecture aims to remove data bottlenecks and enable fast access to structured and semi-structured data, with additional features like always-on data ingestion and in-database AI and machine learning with OcientML.
Amount: $49.4M, Round: Series B extension
URL: https://www.techtarget.com/searchbusinessanalytics/news/366573032/Ocient-raises-494M-in-funding-to-fuel-development-growth

Empathy

Description: Empathy has built a platform to help navigate the practical and emotional bereavement process, incorporating a mixture of AI and human guides to assist with various aspects of bereavement care.
Amount: $47M, Round: Series B
URL: https://techcrunch.com/2024/03/12/empathy-berevement-death-ai/

Eye Security

Description: Eye Security, founded by Dutch Intelligence and Security Services experts, offers enterprise-level 24/7 cyber protection, incident response, and cyber insurance. Trusted by diverse sectors, the company aims to expand its presence in Europe, addressing the cybersecurity needs of mid-market businesses.
Amount: €36M, Round: Series B
URL: https://www.eye.security/press/eye-security-secures-36-million-in-series-b-funding

Nanonets

Description: Nanonets is a startup using AI to automate back-office processes, primarily targeting the financial services sector. Its AI platform aims to improve the efficiency of processes involving unstructured data like invoices, receipts, and purchase orders.
Amount: $29 million, Round: Series B
URL: https://techcrunch.com/2024/03/12/nanonets-funding-accel-india/

Adaptive ML

Description: Adaptive ML is a Paris, France- and New York-based startup enabling companies to perpetually improve their generative AI models based on users’ interactions. The platform abstracts away technical details around fine-tuning and reinforcement learning, empowering enterprises to leverage ever-learning generative models for intuitive experiences in customer support, software development, and deep research efforts.
Amount: $20M, Round: Seed
URL: https://www.finsmes.com/2024/03/adaptive-ml-raises-20m-in-seed-funding.html

Tavus

Description: Tavus is a generative AI startup that helps companies create digital replicas of individuals for automated personalized video campaigns. It uses generative AI to power personalized videos with voice and face cloning, enabling sales and marketing teams to send personalized videos to prospects at scale.
Amount: $18M, Round: Series A
URL: https://techcrunch.com/2024/03/12/generative-ai-video-startup-tavus-raises-18m-to-bring-face-and-voice-cloning-to-any-app/

Neural Propulsion Systems

Description: Neural Propulsion Systems (NPS) pioneers software-defined radar technology, enhancing roadway safety and defense radar systems. Their Atomic Sensing Platform revolutionizes radar resolution and precision, achieving over 10x performance enhancements in the $28 billion radar market.
Amount: $17.5M, Round: Series B
URL: https://www.prnewswire.com/news-releases/nps-completes-17-5m-series-b-funding-for-software-defined-radar-led-by-cota-capital-with-key-strategic-investment-from-gm-ventures-and-rtx-ventures-302082398.html

Axion Ray

Description: Axion Ray is a NYC-based provider of an AI-powered observability command center for global manufacturing companies, enabling early detection and resolution of quality issues. Its customer base includes Fortune 500 manufacturers across various industries like electronics, automotive, and medical devices.
Amount: $17.5M, Round: Series A
URL: https://www.finsmes.com/2024/03/axion-ray-announces-17-5m-in-series-a-funding.html

Axion Ray

Description: Axion Ray is a company creating an AI-powered platform to predict product failures by taking in signals from field service reports to sensor readings and correlating these signals along with geolocation and other data.
Amount: $17.5 million, Round: Series A
URL: https://techcrunch.com/2024/03/12/axion-rays-ai-attempts-to-detect-product-flaws-to-prevent-recalls/

Kaedim

Description: Kaedim provides AI-powered tools for 3D asset content creation, offering a new marketplace with 10,000 pre-generated assets. The platform aims to enhance artists' workflows with solutions like 3D workflow integration, smart add-ons, and custom asset generation services.
Amount: $15M, Round: Series A
URL: https://venturebeat.com/games/kaedim-raises-15m-to-fund-ai-based-3d-asset-creation-solutions/

DBOS

Description: DBOS is a database oriented operating system that runs OS services on top of distributed databases, offering a scalable, fault-tolerant, and cyber-resilient foundation for cloud-native applications. Their first product, DBOS Cloud, is a FaaS platform simplifying development, deployment, and operations, prioritizing cybersecurity and cyber-resilience.
Amount: $8.5M, Round: Seed
URL: https://sdtimes.com/cloud/dbos-announces-faas-platform-dbos-cloud-and-8-5-million-in-seed-funding/

Polar Signals

Description: Polar Signals provides a continuous profiling platform, Polar Signals Cloud, to optimize application performance by identifying excess memory and processor resource usage. The platform integrates with applications seamlessly and offers AI-driven performance optimization suggestions, attracting notable customers like Canonical Ltd.
Amount: $6.8M, Round: extension to seed round
URL: https://siliconangle.com/2024/03/12/polar-signals-reels-6-8m-ease-application-performance-analysis/

Nijta

Description: Nijta is a French startup that provides AI-powered speech anonymization technology to clients in order to help them comply with privacy requirements. The company was founded in Lille, France, by Indian CEO Brij Srivastava, who moved there for his PhD at Inria. Nijta has received €2 million in funding from sources including French deep tech VC fund Elaia and Lille-based investment firm Finovam Gestion. The company's primary market is Europe due to the strong data privacy law GDPR.
Amount: €2 million, Round:
URL: https://techcrunch.com/2024/03/11/nijta-voice-privacy-ai/

Products

Top Products

PitchBob.io AI pitch deck generator & startup co-pilot
Innovative AI assistant transforming ideas into actionable plans for entrepreneurs.

Source: https://www.producthunt.com/r/SU6TC7S7TRLAUW?utm_campaign=producthunt-api&utm_medium=api-v2&utm_source=Application%3A+info.ai+%28ID%3A+116042%29

Tavus for Developers Beta Build human-like replicas and text-to-video with APIs
Revolutionizes content creation with human-like replicas and text-to-video conversion.

Source: https://www.producthunt.com/r/AZL62SSRQ3WQEJ?utm_campaign=producthunt-api&utm_medium=api-v2&utm_source=Application%3A+info.ai+%28ID%3A+116042%29

Codifyer Generate your whole database code in minutes
Automates and optimizes database creation with AI, enhancing developer productivity.

Source: https://www.producthunt.com/r/QEZRSPG7OAT3PZ?utm_campaign=producthunt-api&utm_medium=api-v2&utm_source=Application%3A+info.ai+%28ID%3A+116042%29

Mystic BYOC Run any ML model in your AWS/Azure/GCP account
Empowers ML deployment with serverless experience and cost control in your cloud.

Source: https://www.producthunt.com/r/WDL2TB37CHE42Y?utm_campaign=producthunt-api&utm_medium=api-v2&utm_source=Application%3A+info.ai+%28ID%3A+116042%29

DubVid Translate your videos in any language with just 1-click
Breaks language barriers with 1-click video translation, extending global reach.

Source: https://www.producthunt.com/r/TI5PRQP2Z5456D?utm_campaign=producthunt-api&utm_medium=api-v2&utm_source=Application%3A+info.ai+%28ID%3A+116042%29

Research

Stealing Part of a Production Language Model

Authors: Nicholas Carlini, Daniel Paleka, Krishnamurthy (Dj) Dvijotham, Thomas Steinke, Jonathan Hayase, A. Feder Cooper, Katherine Lee, Matthew Jagielski, Milad Nasr, Arthur Conmy, Eric Wallace, David Rolnick, Florian Tramèr
Institution: Google DeepMind, ETH Zurich, University of Washington, OpenAI, McGill University
This study presents a way to extract detailed information from advanced AI language models like ChatGPT, using a method that costs very little. It successfully reveals the inner dimensions of these models and suggests how to protect them against such attacks. The findings highlight the need for better security measures.
Link: https://arxiv.org/abs/2403.06634

Adding NVMe SSDs to Enable and Accelerate 100B Model Fine-tuning on a Single GPU

Authors: Changyue Liao, Mo Sun, Zihan Yang, Kaiqi Chen, Binhang Yuan, Fei Wu, Zeke Wang
Institution: Zhejiang University, China, HKUST, China
This study introduces Fuyou, a cost-effective training framework that allows for the fine-tuning of huge 100B models on low-end GPUs and servers with limited memory. By optimizing the interaction between SSD storage and CPU, Fuyou maximizes GPU efficiency, enabling the training of models up to 175B parameters on consumer-grade hardware, significantly outperforming existing methods.
Link: https://arxiv.org/abs/2403.06504

V3D: Video Diffusion Models are Effective 3D Generators

Authors: Zilong Chen, Yikai Wang, Feng Wang, Zhengyi Wang, Huaping Liu
Institution: Tsinghua University, ShengShu
This study introduces V3D, a method leveraging video diffusion models to generate high-quality, detailed 3D objects and scenes from a single image. By incorporating geometrical consistency, it ensures the generated 3D representations are realistic from all angles. Extensive testing shows its superiority in producing detailed and consistent 3D models quickly and efficiently. The method also allows for scene-level view synthesis with minimal input, offering potential in various applications.
Link: https://arxiv.org/abs/2403.06738