New and recently funded AI Startups
1
Country: USA
DataAnnotation Tech (a subsidiary of Surge AI) is a platform that specializes in recruiting more-or-less qualified experts to label data for training AI models remotely. Freelancers perform tasks such as image and video labeling, fact-checking and suggesting best responses for chatbots, audio transcription, writing text and code. Pay depends on the complexity of the task and starts at $20 per hour. The platform allows participants to choose projects and working hours. Before contractors can receive tasks, they must create account and complete screening tests. The platform is often associated with scam because freelancers sell their accounts or build a network of subcontractors to work through their own accounts.
DataAnnotation Tech (a subsidiary of Surge AI) is a platform that specializes in recruiting more-or-less qualified experts to label data for training AI models remotely. Freelancers perform tasks such as image and video labeling, fact-checking and suggesting best responses for chatbots, audio transcription, writing text and code. Pay depends on the complexity of the task and starts at $20 per hour. The platform allows participants to choose projects and working hours. Before contractors can receive tasks, they must create account and complete screening tests. The platform is often associated with scam because freelancers sell their accounts or build a network of subcontractors to work through their own accounts.
2
Country: USA | Funding: $337M
Fal is a cloud platform that provides developers with a variety of generative models for image, video, 3D and audio in a single location. Specifically, Fal provides an infrastructure layer for multimodal AI for Adobe, Shopify, Canva and Quora. Developers are provided with a library of over 600 ready-to-use models (including Flux, Nano Banana, Kling Video and Veo) via the API and H100, H200 and B200 virtual machines, as well as dedicated clusters with Fal compute resources for instant inference and rapid scaling from zero to thousands of GPUs. The platform includes tools for model rapid deployment, testing, production deployment and monitoring. Enterprises are offered private model hosting and model co-development options.
Fal is a cloud platform that provides developers with a variety of generative models for image, video, 3D and audio in a single location. Specifically, Fal provides an infrastructure layer for multimodal AI for Adobe, Shopify, Canva and Quora. Developers are provided with a library of over 600 ready-to-use models (including Flux, Nano Banana, Kling Video and Veo) via the API and H100, H200 and B200 virtual machines, as well as dedicated clusters with Fal compute resources for instant inference and rapid scaling from zero to thousands of GPUs. The platform includes tools for model rapid deployment, testing, production deployment and monitoring. Enterprises are offered private model hosting and model co-development options.
3
Country: USA | Funding: $9.3M
Onton is developing AI-chat-based search engine for online furniture stores. The company uses so-called neurosymbolic architecture to overcome the hallucination problems inherent in LLM and provide higher-quality, logical search results. The startup's model can also be trained using real-world information that doesn't necessarily need to be included in product descriptions. For example, if you search for pet-friendly furniture, the model will learn and start return only furniture that is stain- and scratch-resistant. You can also upload an image to Onton's chat to generate a desired design for your home or office, and Onton will then select furniture based on it. The service also offers an infinite image generation canvas, where you can add existing images along with products you've found for inspiration.
Onton is developing AI-chat-based search engine for online furniture stores. The company uses so-called neurosymbolic architecture to overcome the hallucination problems inherent in LLM and provide higher-quality, logical search results. The startup's model can also be trained using real-world information that doesn't necessarily need to be included in product descriptions. For example, if you search for pet-friendly furniture, the model will learn and start return only furniture that is stain- and scratch-resistant. You can also upload an image to Onton's chat to generate a desired design for your home or office, and Onton will then select furniture based on it. The service also offers an infinite image generation canvas, where you can add existing images along with products you've found for inspiration.
4
Country: Finland | Funding: €100M
NestAI is a physical AI lab that supports Finnish Defence Forces in adopting AI. It also has a partnership with hardware giant Nokia to build AI products for defense applications using large language models and related technology for robotics and other real-world applications. NestAI's systems perceive environment with sensors and cameras, think or make decisions (via LLM models), act in physical space via robots, manipulators, drones.
NestAI is a physical AI lab that supports Finnish Defence Forces in adopting AI. It also has a partnership with hardware giant Nokia to build AI products for defense applications using large language models and related technology for robotics and other real-world applications. NestAI's systems perceive environment with sensors and cameras, think or make decisions (via LLM models), act in physical space via robots, manipulators, drones.
5
Country: USA | Funding: $25M
Spline offers collaborative 3D design platform that allows to create and manage 3D models and environments. It is a completely cloud-based and works in a browser (does not require hardware/software installation). It uses AI to create any 3D objects and 3D video scenes from text prompts and images. After generation, you can continue editing objects and scenes in the online Spline editor. An additional tool allows you to transfer 3D-model styles. The service allows to easily export AI content or embed into websites or mobile applications with the help of a code fragment. Spline offers a consumer (paid) chat for 3D generation as well as an enterprise version that meets all security and data privacy requirements
Spline offers collaborative 3D design platform that allows to create and manage 3D models and environments. It is a completely cloud-based and works in a browser (does not require hardware/software installation). It uses AI to create any 3D objects and 3D video scenes from text prompts and images. After generation, you can continue editing objects and scenes in the online Spline editor. An additional tool allows you to transfer 3D-model styles. The service allows to easily export AI content or embed into websites or mobile applications with the help of a code fragment. Spline offers a consumer (paid) chat for 3D generation as well as an enterprise version that meets all security and data privacy requirements
6
Country: USA | Funding: $27M
Odyssey is an AI lab focused on creating universal world models, which the company calls new form of audiovisual intelligence. These models are to form the basis of the next generation of games, films, education content, training simulations and advertising. These models don't just generate video - they enable interaction with the 3D world i.e. creation of interactive videos. The company achieves this through a new multi-stage training pipeline that transforms the model into a causal behavioral video model that reacts to actions in real time and continuously responds to input. As you play the video, you shape it in real time using natural text prompts, similar to communicating with a language model.
Odyssey is an AI lab focused on creating universal world models, which the company calls new form of audiovisual intelligence. These models are to form the basis of the next generation of games, films, education content, training simulations and advertising. These models don't just generate video - they enable interaction with the 3D world i.e. creation of interactive videos. The company achieves this through a new multi-stage training pipeline that transforms the model into a causal behavioral video model that reacts to actions in real time and continuously responds to input. As you play the video, you shape it in real time using natural text prompts, similar to communicating with a language model.
7
Country: USA | Funding: $1B
Harvey develops AI assistant for legal/knowledge workers. It allows professional service providers, and Fortune 500 companies delegate complex legal tasks in natural language. It's secure cloud service allows to upload, store and analyze thousands of documents to use it as a context. This data is enriched by 150 legal data sources across the globe. Lawyers can design and deploy multi-model agents tailored to firm’s expertise and deliver consistent, high-quality results at scale to enhance drafting and editing, email management and knowledge sharing in the Microsoft office tools. Harvey was one of OpenAI Startup Fund’s first investments, so it primary uses its API.
Harvey develops AI assistant for legal/knowledge workers. It allows professional service providers, and Fortune 500 companies delegate complex legal tasks in natural language. It's secure cloud service allows to upload, store and analyze thousands of documents to use it as a context. This data is enriched by 150 legal data sources across the globe. Lawyers can design and deploy multi-model agents tailored to firm’s expertise and deliver consistent, high-quality results at scale to enhance drafting and editing, email management and knowledge sharing in the Microsoft office tools. Harvey was one of OpenAI Startup Fund’s first investments, so it primary uses its API.
8
Country: USA | Funding: $41.6M
Micro1 helps AI development companies find and manage contractors for data labeling and training. Micro1 partners with leading AI labs, including Microsoft, who are seeking to improve large-scale language models using post-training and reinforcement learning. The company also helps evaluate enterprise AI agents for internal workflows, operations support, finance and industry-specific tasks - with the help of subject-matter experts. Micro1 also enables robotics pre-training, which requires high-quality, human-generated demonstrations of everyday physical tasks. It's building the world's largest robotics pre-training dataset by collecting demonstrations from hundreds of generalists who record interactions with objects in their homes.
Micro1 helps AI development companies find and manage contractors for data labeling and training. Micro1 partners with leading AI labs, including Microsoft, who are seeking to improve large-scale language models using post-training and reinforcement learning. The company also helps evaluate enterprise AI agents for internal workflows, operations support, finance and industry-specific tasks - with the help of subject-matter experts. Micro1 also enables robotics pre-training, which requires high-quality, human-generated demonstrations of everyday physical tasks. It's building the world's largest robotics pre-training dataset by collecting demonstrations from hundreds of generalists who record interactions with objects in their homes.
9
Country: USA | Funding: $133.7M
General Intuition is a spinoff of Medal, a platform for uploading and sharing videogame clips. The startup uses this vast collection of game videos to train and create base models and AI agents capable of understanding how objects and entities move through space and time. General Intuition's model is capable of understanding environments it hasn't been trained on and correctly predicts actions within them. It does this solely based on visual input: the agents see only what the human player sees and navigate space following the controller's directions. According to the company, this approach can be naturally transferred to physical systems such as robotic arms, drones and autonomous vehicles, which people often control with video game controllers.
General Intuition is a spinoff of Medal, a platform for uploading and sharing videogame clips. The startup uses this vast collection of game videos to train and create base models and AI agents capable of understanding how objects and entities move through space and time. General Intuition's model is capable of understanding environments it hasn't been trained on and correctly predicts actions within them. It does this solely based on visual input: the agents see only what the human player sees and navigate space following the controller's directions. According to the company, this approach can be naturally transferred to physical systems such as robotic arms, drones and autonomous vehicles, which people often control with video game controllers.
10
Country: USA | Funding: $252M
Augment creates an AI platform for developers that includes chat, AI agent and Next Edit (code completion tool). The company claims that their AI assistants better understand programmers' intent than GitHub Copilot for example. Augment's autonomous software agents can run in various IDEs and in the cloud. The agent has a Memories feature, which is automatically updated during development and persists between conversations to continuously improve code generation, solve problems faster and match developer's style and coding patterns. The model supports MCP link to GitHub, Jira, Confluence, Notion, Linear, Vercel and Cloudflare. Augment's models were trained on publicly available data, some of which may be copyrighted or have a restrictive license.
Augment creates an AI platform for developers that includes chat, AI agent and Next Edit (code completion tool). The company claims that their AI assistants better understand programmers' intent than GitHub Copilot for example. Augment's autonomous software agents can run in various IDEs and in the cloud. The agent has a Memories feature, which is automatically updated during development and persists between conversations to continuously improve code generation, solve problems faster and match developer's style and coding patterns. The model supports MCP link to GitHub, Jira, Confluence, Notion, Linear, Vercel and Cloudflare. Augment's models were trained on publicly available data, some of which may be copyrighted or have a restrictive license.
