Station F: Europe's AI Startup Launchpad
Station F, a prominent startup hub in Paris, is intensifying its focus on AI by launching the second cohort of its F/ai accelerator program. This initiative aims to propel AI startups from early-stage products to revenue generation rapidly. The program leverages Station F's extensive network, including relationships with major tech companies and VCs, to provide startups with crucial connections and resources.
Nearly all of Station F's 'Future 40' selections in 2024 incorporated AI, highlighting the hub's strategic position in fostering Europe's burgeoning AI ecosystem. The accelerator's success is evident in its ability to attract significant industry backing and in the achievements of its participating startups, demonstrating its role as a vital launchpad for AI innovation in Europe.
Station F, a prominent startup hub in Paris, is intensifying its focus on AI by launching the second cohort of its F/ai accelerator program. This initiative aims to propel AI startups from early-stage products to revenue generation rapidly. The program leverages Station F's extensive network, including relationships with major tech companies and VCs, to provide startups with crucial connections and resources.
Nearly all of Station F's 'Future 40' selections in 2024 incorporated AI, highlighting the hub's strategic position in fostering Europe's burgeoning AI ecosystem. The accelerator's success is evident in its ability to attract significant industry backing and in the achievements of its participating startups, demonstrating its role as a vital launchpad for AI innovation in Europe.
Google AI Training: How to Opt Out
Google has updated its privacy settings to allow the use of user media, such as images and audio, for AI model development. This change was communicated via an email update in June and impacts various Google services including Search, Maps, and Translate.
This move reflects a broader industry trend where tech companies are leveraging user-generated content to enhance AI capabilities. Users can opt out of this data usage by adjusting their Search Services History and Personalized Recommendations settings on Google's privacy pages. #AI #Google #Privacy #Tech
Google has updated its privacy settings to allow the use of user media, such as images and audio, for AI model development. This change was communicated via an email update in June and impacts various Google services including Search, Maps, and Translate.
This move reflects a broader industry trend where tech companies are leveraging user-generated content to enhance AI capabilities. Users can opt out of this data usage by adjusting their Search Services History and Personalized Recommendations settings on Google's privacy pages. #AI #Google #Privacy #Tech
Savi App Fights Evolving AI Scams
Savi Security has launched a new app designed to combat the growing threat of sophisticated AI-generated scams. The company, which raised $7 million in seed funding, offers real-time protection against deceptive texts, emails, and phone calls. Inspired by a family encounter with a convincing AI voice-spoofing scam, Savi aims to equip consumers with AI-powered tools to detect and prevent fraud. The app features live call monitoring and family-wide protection for $8/month.
This development is crucial as AI lowers the barrier for cybercriminals, making advanced scams accessible to a wider audience. Savi's approach mirrors the adversarial nature of AI, using similar technology to defend against these emerging threats. #AISecurity #ScamPrevention #TechInnovation
Savi Security has launched a new app designed to combat the growing threat of sophisticated AI-generated scams. The company, which raised $7 million in seed funding, offers real-time protection against deceptive texts, emails, and phone calls. Inspired by a family encounter with a convincing AI voice-spoofing scam, Savi aims to equip consumers with AI-powered tools to detect and prevent fraud. The app features live call monitoring and family-wide protection for $8/month.
This development is crucial as AI lowers the barrier for cybercriminals, making advanced scams accessible to a wider audience. Savi's approach mirrors the adversarial nature of AI, using similar technology to defend against these emerging threats. #AISecurity #ScamPrevention #TechInnovation
US Autonomous Vehicles Deployed in Ukraine
Over 100 of Forterra's self-driving ATVs have been operating in conflict zones in Ukraine for nine months, marking a significant deployment of autonomous ground vehicles in combat. These gas-powered Lancer vehicles, capable of carrying 750 kg, are crucial for logistics, resupply, and casualty evacuation, especially in areas made dangerous by constant drone surveillance. While primarily teleoperated due to current AI limitations in threat identification, these vehicles represent a major step in integrating autonomous technology into modern warfare, offering valuable combat lessons for future defense applications.
#AI #AutonomousVehicles #DefenseTech #Ukraine
Over 100 of Forterra's self-driving ATVs have been operating in conflict zones in Ukraine for nine months, marking a significant deployment of autonomous ground vehicles in combat. These gas-powered Lancer vehicles, capable of carrying 750 kg, are crucial for logistics, resupply, and casualty evacuation, especially in areas made dangerous by constant drone surveillance. While primarily teleoperated due to current AI limitations in threat identification, these vehicles represent a major step in integrating autonomous technology into modern warfare, offering valuable combat lessons for future defense applications.
#AI #AutonomousVehicles #DefenseTech #Ukraine
AI Ransomware Attack Still Needs Human Touch
A recent report highlighted the first known instance of "agentic ransomware" named JadePuffer, where an AI agent handled a cyberattack from start to finish. However, clarifications reveal that a human was still essential for initiating the attack, selecting the victim, and providing necessary credentials. While the AI autonomously navigated the network, exploited vulnerabilities, and encrypted files, the initial setup and crucial stolen credentials still required human intervention.
This development, though not fully autonomous, is significant as it demonstrates AI's growing capability in executing complex cyber threats. The AI's ability to adapt and proceed without direct human oversight during the attack phase marks a concerning advancement in cybersecurity. #AI #Ransomware #Cybersecurity
A recent report highlighted the first known instance of "agentic ransomware" named JadePuffer, where an AI agent handled a cyberattack from start to finish. However, clarifications reveal that a human was still essential for initiating the attack, selecting the victim, and providing necessary credentials. While the AI autonomously navigated the network, exploited vulnerabilities, and encrypted files, the initial setup and crucial stolen credentials still required human intervention.
This development, though not fully autonomous, is significant as it demonstrates AI's growing capability in executing complex cyber threats. The AI's ability to adapt and proceed without direct human oversight during the attack phase marks a concerning advancement in cybersecurity. #AI #Ransomware #Cybersecurity
Claude Cowork Goes Mobile and Web
Anthropic's AI assistant for general knowledge work, Claude Cowork, is expanding beyond its desktop app to become accessible on web and mobile for Max subscribers. This move signifies Anthropic's ambition to position Cowork not just as a coding tool, but as an adaptable AI coworker that can operate seamlessly across devices, manage background tasks, and prompt user input when necessary. This expansion into everyday work surfaces mirrors similar strategies from other AI companies, suggesting a shift towards AI agents integrated into users' daily workflows rather than solely existing as standalone chatbots.
#AI #Claude #Anthropic #Tech
Anthropic's AI assistant for general knowledge work, Claude Cowork, is expanding beyond its desktop app to become accessible on web and mobile for Max subscribers. This move signifies Anthropic's ambition to position Cowork not just as a coding tool, but as an adaptable AI coworker that can operate seamlessly across devices, manage background tasks, and prompt user input when necessary. This expansion into everyday work surfaces mirrors similar strategies from other AI companies, suggesting a shift towards AI agents integrated into users' daily workflows rather than solely existing as standalone chatbots.
#AI #Claude #Anthropic #Tech
ZML Releases Free AI Inference Software
French AI startup ZML, endorsed by Yann LeCun, has launched ZML/LLMD, a free inference server designed to optimize large language model performance across a wide range of hardware, including Nvidia, AMD, Google TPUs, Apple Metal, and Intel Arc. This move aims to break down existing silos in AI chip utilization, potentially reducing costs and increasing efficiency for enterprises.
The software's goal is to enable various open-source LLMs to run at maximum speed on diverse chips, challenging Nvidia's dominance and offering more flexibility in AI deployments. This is particularly significant amid growing concerns about the rising costs associated with AI. ZML plans to iterate on the product and explore monetization strategies later.
#AI #Inference #ZML #TechInnovation
French AI startup ZML, endorsed by Yann LeCun, has launched ZML/LLMD, a free inference server designed to optimize large language model performance across a wide range of hardware, including Nvidia, AMD, Google TPUs, Apple Metal, and Intel Arc. This move aims to break down existing silos in AI chip utilization, potentially reducing costs and increasing efficiency for enterprises.
The software's goal is to enable various open-source LLMs to run at maximum speed on diverse chips, challenging Nvidia's dominance and offering more flexibility in AI deployments. This is particularly significant amid growing concerns about the rising costs associated with AI. ZML plans to iterate on the product and explore monetization strategies later.
#AI #Inference #ZML #TechInnovation
SambaNova Secures $1 Billion at $11 Billion Valuation
AI chip maker SambaNova Systems has successfully raised $1 billion in a Series F funding round, achieving a valuation of $11 billion. This significant investment, led by General Atlantic, comes just five months after their previous $350 million Series E round. The company also announced a strategic partnership with JPMorgan Chase to provide on-premises AI inference infrastructure, signaling a growing trend among financial institutions to move away from cloud-dependent solutions.
This funding and partnership highlight SambaNova's strong position in the competitive AI hardware market. The company's focus on specialized chips for large-scale AI models, coupled with high-profile client wins like JPMorgan Chase, demonstrates the increasing demand for efficient and secure on-premises AI solutions. The capital raised will be used to scale operations and strengthen its supply chain to meet this demand.
#AI #Funding #SambaNova #AIChips
AI chip maker SambaNova Systems has successfully raised $1 billion in a Series F funding round, achieving a valuation of $11 billion. This significant investment, led by General Atlantic, comes just five months after their previous $350 million Series E round. The company also announced a strategic partnership with JPMorgan Chase to provide on-premises AI inference infrastructure, signaling a growing trend among financial institutions to move away from cloud-dependent solutions.
This funding and partnership highlight SambaNova's strong position in the competitive AI hardware market. The company's focus on specialized chips for large-scale AI models, coupled with high-profile client wins like JPMorgan Chase, demonstrates the increasing demand for efficient and secure on-premises AI solutions. The capital raised will be used to scale operations and strengthen its supply chain to meet this demand.
#AI #Funding #SambaNova #AIChips
Robotics Poised for ChatGPT-like Leap
Startup General Intuition is pioneering a new approach to AI, believing that video game data can train foundation models for physical AI. By leveraging millions of hours of gameplay, including controller inputs, they aim to develop models with human-like spatial-temporal reasoning. This method contrasts with traditional approaches that rely on vast, specialized real-world datasets.
This innovation is significant because it could dramatically accelerate the development of more capable and adaptable robots. General Intuition's model has already shown promise by powering a quadrupedal robot after minimal real-world fine-tuning, suggesting a future where robots can learn and adapt more efficiently. The company's vision is to provide the foundational AI for other robotics firms, rather than building robots themselves.
#AI #Robotics #FoundationModels #GeneralIntuition
Startup General Intuition is pioneering a new approach to AI, believing that video game data can train foundation models for physical AI. By leveraging millions of hours of gameplay, including controller inputs, they aim to develop models with human-like spatial-temporal reasoning. This method contrasts with traditional approaches that rely on vast, specialized real-world datasets.
This innovation is significant because it could dramatically accelerate the development of more capable and adaptable robots. General Intuition's model has already shown promise by powering a quadrupedal robot after minimal real-world fine-tuning, suggesting a future where robots can learn and adapt more efficiently. The company's vision is to provide the foundational AI for other robotics firms, rather than building robots themselves.
#AI #Robotics #FoundationModels #GeneralIntuition
Google's SynthID Debunks McConnell Deepfake
Google's AI detection system, SynthID, has successfully debunked a widely circulated hoax image of Senator Mitch McConnell. The image, appearing to show the senator in distress, was identified as AI-generated due to the invisible watermark embedded by SynthID. This marks a significant win for anti-deepfake technology, demonstrating the system's effectiveness even after images are shared across multiple platforms.
SynthID, launched in 2025, works by embedding an imperceptible digital signature into AI-generated images created by participating models like Gemini and OpenAI's. This watermark persists through various image manipulations, allowing tools like SynthID or OpenAI's verification service to detect AI origins. The system's reliability in this high-profile case highlights its growing importance in combating misinformation.
Google's AI detection system, SynthID, has successfully debunked a widely circulated hoax image of Senator Mitch McConnell. The image, appearing to show the senator in distress, was identified as AI-generated due to the invisible watermark embedded by SynthID. This marks a significant win for anti-deepfake technology, demonstrating the system's effectiveness even after images are shared across multiple platforms.
SynthID, launched in 2025, works by embedding an imperceptible digital signature into AI-generated images created by participating models like Gemini and OpenAI's. This watermark persists through various image manipulations, allowing tools like SynthID or OpenAI's verification service to detect AI origins. The system's reliability in this high-profile case highlights its growing importance in combating misinformation.
SpaceXAI Unveils Grok 4.5: 'Opus-Class' AI Model Launched
SpaceXAI has launched its latest AI model, Grok 4.5, positioning it as a direct competitor to high-end models like Anthropic's Opus. Elon Musk described Grok 4.5 as an 'Opus-class model' that is faster, more token-efficient, and lower cost, aiming to tackle a wide range of tasks including coding, office work, research, and writing. The company highlights its model's 'twice greater token efficiency' as a significant advantage in a market concerned with rising AI costs.
This release intensifies the competition in the advanced AI sector, particularly with OpenAI's upcoming GPT 5.6. Grok 4.5's competitive pricing and performance claims could significantly impact the AI market by offering a powerful yet more accessible alternative.
#AI #Grok #SpaceXAI #LLM
SpaceXAI has launched its latest AI model, Grok 4.5, positioning it as a direct competitor to high-end models like Anthropic's Opus. Elon Musk described Grok 4.5 as an 'Opus-class model' that is faster, more token-efficient, and lower cost, aiming to tackle a wide range of tasks including coding, office work, research, and writing. The company highlights its model's 'twice greater token efficiency' as a significant advantage in a market concerned with rising AI costs.
This release intensifies the competition in the advanced AI sector, particularly with OpenAI's upcoming GPT 5.6. Grok 4.5's competitive pricing and performance claims could significantly impact the AI market by offering a powerful yet more accessible alternative.
#AI #Grok #SpaceXAI #LLM
Google Photos Introduces AI Video Remix Tool
Google Photos is rolling out a new 'Video Remix' feature, powered by Gemini Omni, allowing users to easily edit and transform videos with AI. This feature aims to simplify video editing, making it accessible within the Google Photos app. Users can apply effects like cinematic relighting, background swaps, and artistic styles. The tool is becoming available to eligible Google subscribers in select regions.
This integration of generative AI into a popular consumer app highlights Google's strategy to enhance user experience and compete in the AI-driven market. It democratizes video editing, enabling users to create share-worthy content with minimal effort.
#AI #GooglePhotos #GeminiOmni #VideoEditing
Google Photos is rolling out a new 'Video Remix' feature, powered by Gemini Omni, allowing users to easily edit and transform videos with AI. This feature aims to simplify video editing, making it accessible within the Google Photos app. Users can apply effects like cinematic relighting, background swaps, and artistic styles. The tool is becoming available to eligible Google subscribers in select regions.
This integration of generative AI into a popular consumer app highlights Google's strategy to enhance user experience and compete in the AI-driven market. It democratizes video editing, enabling users to create share-worthy content with minimal effort.
#AI #GooglePhotos #GeminiOmni #VideoEditing
Gaming Data for AI Training
Startups like General Intuition are betting that video game data, rather than internet text, is key to achieving artificial general intelligence (AGI). While large language models excel at text, they struggle with understanding physical interactions in space and time, a crucial element for generalized intelligence.
General Intuition, backed by notable investors like Jeff Bezos, Coatue, and Eric Schmidt, has raised $320 million. They believe training world models on gaming data will be a significant leap in physical AI development. The company originated from the gaming platform Medal TV.
Startups like General Intuition are betting that video game data, rather than internet text, is key to achieving artificial general intelligence (AGI). While large language models excel at text, they struggle with understanding physical interactions in space and time, a crucial element for generalized intelligence.
General Intuition, backed by notable investors like Jeff Bezos, Coatue, and Eric Schmidt, has raised $320 million. They believe training world models on gaming data will be a significant leap in physical AI development. The company originated from the gaming platform Medal TV.
Lovable Eyes $13.2B Valuation
Swedish AI startup Lovable is reportedly in talks to double its valuation to $13.2 billion, aiming to raise $300 million in a new funding round led by Menlo Ventures. This move comes after the company achieved a $6.6 billion valuation just six months ago and reached an annualized revenue run rate of $500 million in June.
Lovable's success highlights the explosive growth and potential of 'vibe coding,' a popular AI application allowing users to build software through descriptive language. The company serves individual creators and large enterprises alike, including Workday, Asana, and Nvidia. This significant valuation increase positions Lovable among other high-profile AI startups like Replit and Factory, underscoring the intense interest and investment in AI-powered development tools.
Swedish AI startup Lovable is reportedly in talks to double its valuation to $13.2 billion, aiming to raise $300 million in a new funding round led by Menlo Ventures. This move comes after the company achieved a $6.6 billion valuation just six months ago and reached an annualized revenue run rate of $500 million in June.
Lovable's success highlights the explosive growth and potential of 'vibe coding,' a popular AI application allowing users to build software through descriptive language. The company serves individual creators and large enterprises alike, including Workday, Asana, and Nvidia. This significant valuation increase positions Lovable among other high-profile AI startups like Replit and Factory, underscoring the intense interest and investment in AI-powered development tools.
Meta AI Glasses: Privacy vs. Progress
Meta is attempting to address privacy concerns surrounding its AI-powered glasses by implementing a new safeguard: disabling the camera if the recording indicator light is tampered with. This move comes amid growing public unease about the potential for these devices to be used for surveillance.
However, this effort to mitigate perceived creepiness is juxtaposed with Meta's broader AI strategy, which involves collecting and utilizing user data extensively. This includes training AI on user images, using personal content for AI features by default, and exploring continuous recording and facial recognition. The company's own privacy policy states that images shared with Meta AI can be used for training, raising further questions about user data control and privacy.
#MetaAI #SmartGlasses #Privacy #AIConserns
Meta is attempting to address privacy concerns surrounding its AI-powered glasses by implementing a new safeguard: disabling the camera if the recording indicator light is tampered with. This move comes amid growing public unease about the potential for these devices to be used for surveillance.
However, this effort to mitigate perceived creepiness is juxtaposed with Meta's broader AI strategy, which involves collecting and utilizing user data extensively. This includes training AI on user images, using personal content for AI features by default, and exploring continuous recording and facial recognition. The company's own privacy policy states that images shared with Meta AI can be used for training, raising further questions about user data control and privacy.
#MetaAI #SmartGlasses #Privacy #AIConserns
OpenAI Designates GPT 5.6 as Preferred Model for Microsoft Copilot
Amidst speculation about a potential rift, OpenAI has announced that its new GPT 5.6 model will be the 'preferred model' for Microsoft's 365 Copilot. This move aims to power productivity apps like Word and Excel, reinforcing the partnership between the two tech giants. While Microsoft has been exploring in-house AI models to cut costs, OpenAI's designation suggests a continued, albeit perhaps adjusted, collaboration.
#AI #Microsoft #OpenAI #Copilot
Amidst speculation about a potential rift, OpenAI has announced that its new GPT 5.6 model will be the 'preferred model' for Microsoft's 365 Copilot. This move aims to power productivity apps like Word and Excel, reinforcing the partnership between the two tech giants. While Microsoft has been exploring in-house AI models to cut costs, OpenAI's designation suggests a continued, albeit perhaps adjusted, collaboration.
#AI #Microsoft #OpenAI #Copilot
OpenAI's No. 2 Executive Steps Down
Fidji Simo, OpenAI's second-in-command, is transitioning from her full-time role to a part-time advisory position due to ongoing health challenges. Simo, who joined OpenAI in May 2025 as CEO of Applications, was instrumental in consolidating business and product operations. Her departure creates a significant leadership void as OpenAI prepares for a potential IPO and intensifies its efforts to compete in the enterprise AI market.
#OpenAI #AILeadership #TechNews #Layoffs
Fidji Simo, OpenAI's second-in-command, is transitioning from her full-time role to a part-time advisory position due to ongoing health challenges. Simo, who joined OpenAI in May 2025 as CEO of Applications, was instrumental in consolidating business and product operations. Her departure creates a significant leadership void as OpenAI prepares for a potential IPO and intensifies its efforts to compete in the enterprise AI market.
#OpenAI #AILeadership #TechNews #Layoffs
Apple Sues OpenAI for Trade Secret Theft
Apple has filed a lawsuit against OpenAI, accusing the AI company and its senior leadership of trade secret theft and breach of contract. The tech giant alleges that former Apple employees now at OpenAI, including Chief Hardware Officer Tang Tan, have misappropriated confidential Apple information, including details about unannounced products and proprietary code. This lawsuit arises amid rumors of OpenAI's development of its own hardware product, which could directly compete with Apple's core business.
The legal action highlights the intense competition and ongoing battles over intellectual property in the rapidly evolving AI and tech landscape. Apple aims to prevent OpenAI from using its trade secrets and to uncover the full extent of the alleged misconduct through legal discovery.
Apple has filed a lawsuit against OpenAI, accusing the AI company and its senior leadership of trade secret theft and breach of contract. The tech giant alleges that former Apple employees now at OpenAI, including Chief Hardware Officer Tang Tan, have misappropriated confidential Apple information, including details about unannounced products and proprietary code. This lawsuit arises amid rumors of OpenAI's development of its own hardware product, which could directly compete with Apple's core business.
The legal action highlights the intense competition and ongoing battles over intellectual property in the rapidly evolving AI and tech landscape. Apple aims to prevent OpenAI from using its trade secrets and to uncover the full extent of the alleged misconduct through legal discovery.
SK Hynix's Landmark US IPO Fuels AI Chip Boom
South Korean memory chip giant SK Hynix has achieved a historic milestone with its U.S. market debut, raising $26.5 billion in the largest-ever IPO by a non-American company. This success is largely attributed to the surging demand for AI chips, particularly High-Bandwidth Memory (HBM), a critical component for AI GPUs, where SK Hynix is a key supplier to NVIDIA. The funds raised will be invested in new manufacturing facilities in South Korea, bolstering the global supply of essential memory chips.
Amidst this significant financial event, U.S. Commerce Secretary Howard Lutnick is actively engaging with SK Hynix and Samsung to encourage the establishment of new semiconductor fabrication plants within the United States. This strategic push aims to diversify the global chip manufacturing landscape and reduce reliance on single regions, further solidifying the AI chip sector's economic and geopolitical importance. #AI #Semiconductors #IPO #SKHynix
South Korean memory chip giant SK Hynix has achieved a historic milestone with its U.S. market debut, raising $26.5 billion in the largest-ever IPO by a non-American company. This success is largely attributed to the surging demand for AI chips, particularly High-Bandwidth Memory (HBM), a critical component for AI GPUs, where SK Hynix is a key supplier to NVIDIA. The funds raised will be invested in new manufacturing facilities in South Korea, bolstering the global supply of essential memory chips.
Amidst this significant financial event, U.S. Commerce Secretary Howard Lutnick is actively engaging with SK Hynix and Samsung to encourage the establishment of new semiconductor fabrication plants within the United States. This strategic push aims to diversify the global chip manufacturing landscape and reduce reliance on single regions, further solidifying the AI chip sector's economic and geopolitical importance. #AI #Semiconductors #IPO #SKHynix
Hugging Face CEO: Companies Shifting to Open Source AI
Hugging Face CEO Clem Delangue highlights a significant trend: companies are increasingly moving from expensive, rented AI APIs to open-source models as they scale. This shift is driven by cost-effectiveness and greater control.
Hugging Face has become a central hub for AI development, akin to a 'GitHub for AI,' with many Fortune 500 companies leveraging its platform for open models and datasets. Delangue emphasizes the importance of the open-source movement in preventing a few large companies from dominating the AI landscape, especially in light of recent debates around AI safety and accessibility.
#AI #OpenSourceAI #HuggingFace #Tech
Hugging Face CEO Clem Delangue highlights a significant trend: companies are increasingly moving from expensive, rented AI APIs to open-source models as they scale. This shift is driven by cost-effectiveness and greater control.
Hugging Face has become a central hub for AI development, akin to a 'GitHub for AI,' with many Fortune 500 companies leveraging its platform for open models and datasets. Delangue emphasizes the importance of the open-source movement in preventing a few large companies from dominating the AI landscape, especially in light of recent debates around AI safety and accessibility.
#AI #OpenSourceAI #HuggingFace #Tech
AI Agent Successfully Runs $100M Fundraise
Startup Lyzr utilized its own AI agent, SivaClaw, to manage its $100 million Series B funding round. The AI reportedly handled investor communications, drafted memos, and even analyzed engagement with presentation slides, showcasing the practical capabilities of their product. This innovative approach secured significant interest without requiring extensive founder travel.
#AI #Startup #Fundraising #AIAgent
Startup Lyzr utilized its own AI agent, SivaClaw, to manage its $100 million Series B funding round. The AI reportedly handled investor communications, drafted memos, and even analyzed engagement with presentation slides, showcasing the practical capabilities of their product. This innovative approach secured significant interest without requiring extensive founder travel.
#AI #Startup #Fundraising #AIAgent