Meet the 40 Startups Named to AWS’ 2025 Generative AI Accelerator

Amazon Web Services (AWS) has selected the startups that will participate in its generative AI accelerator (GAIA). The third cohort comprises a diverse mix of companies from software, life sciences, healthcare, media and entertainment, finance, professional services, manufacturing, cybersecurity, legal, robotics, and education sectors—all with one thing in common: they are utilizing AI.

“This year’s cohort reinforces our mission to help that innovation move faster and deliver real-world impact for customers in every industry,” Sherry Karamdashti, AWS’ general manager and head of startups in North America, says in a release. “We’re removing the barriers and accelerating opportunities so these leaders can grow their world-changing solutions.”

Unlike previous years, AWS has limited the number of participating startups. In 2024, there were 80 enrolled, but now there are 40. The program is also on a truncated timeline, with two fewer weeks. However, AWS’ commitment remains the same, investing up to $1 million in AWS credits to each company, providing technical support, mentorship, and go-to market support, and access to its gen AI technology.

Here are the startups enrolled in GAIA this year (in alphabetical order)

• AI Cube (Brazil): No-code AI automation to streamline tasks and boost workflow efficiency
• Basetwo AI (Canada): AI copilot helping engineers optimize pharmaceutical manufacturing in real time
• Chai Discovery (U.S.): Uses AI to design better molecules and accelerate drug discovery
• Dharma-AI (Brazil): Builds small language models for secure, efficient enterprise automation
• Eloquent AI (U.S.): AI operator automating complex financial operations without APIs or coding
• Exaforce (U.S.): AI-powered SOC platform enhancing cybersecurity detection and response
• Forlex (Brazil): Legal AI agents and multilingual models transforming law firm operations
• Hedra Inc. (U.S.): Generative media platform for end-to-end marketing and content creation
• Hemispheric (Israel): AI foundation for brain-computer interfaces advancing neuroscience
• Hyperbots (India): Agentic AI platform boosting finance team productivity by over 90 percent
• Inception (U.S.): Developing diffusion LLMs to make AI faster and cheaper
• Inephany (U.K.): Optimizes machine learning training speed and cost through AI tools
• Invisible Universe (U.S.): Builds AI-native studios for creators making short-form videos
• Jentic (Ireland): Deploys AI agents in secure sandboxes for enterprise automation
• Lettria (France): No-code AI GraphRAG platform turning unstructured data into insights
• Lisan (UAE): Multilingual AI writing and document generation tools for enterprises
• LlamaIndex (U.S.): Framework for building agentic workflows to extract and synthesize knowledge
• Manifold Bio (U.S.): Uses AI and protein engineering to develop targeted therapeutics.
• Mary Technology (Australia): AI-powered fact management system for law firms
• Mimic Robotics (Switzerland): AI robotics automating complex manual tasks in retail and manufacturing
• NeuBird (U.S.): Autonomous AI system resolving IT incidents rapidly
• Nexxa.ai (U.S.): Specialized AI for heavy industries to accelerate engineering work
• Orakl Oncology (France): AI-powered predictive modeling to improve drug development outcomes
• Pathway (U.S.): Builds real-time learning AI systems to mimic human cognition
• Pluralis Research (Australia): Enables collaborative foundation model training among organizations
• Qomplement (Mexico): Automates back-office tasks across legacy systems using computer-use agents
• Ravenna (U.S.): AI help desk automating IT, HR, and operations support in Slack
• RLWRLD (Korea): Robotics foundation models for industrial automation
• Reevo (U.S.): AI platform unifying B2B sales and marketing operations
• Runloop (U.S.): Developer infrastructure for building AI agent-based software
• SDio (Japan): Transforms long videos into searchable, analyzable intelligence
• Smallest AI (India): Foundational voice AI models for real-time enterprise communication
• Stimuler (India): Voice AI app teaching English fluency to non-native speakers
• Synthera (Colombia): AI simulations for modeling financial markets and macro trends
• SyntheticGestalt (Japan): Molecular-focused foundation AI model accelerating scientific discovery
• Trillion Labs (Korea): Builds large language models for Korean and Asian languages
• Typhoon (Thailand): Open-source multimodal LLMs for Thai developers and businesses
• VidLab7 (Germany): AI demo platform creating interactive sales avatars
• Wondera (U.S.): AI music creation and publishing tools empowering creators

AWS notes that all of these startups are making notable contributions in the field of gen AI, whether it involves fine-tuning open-source models, pre-training foundation models, or providing critical services to the tech stack.

All will be gathering in Seattle, Washington, on October 14 to kick off the 2025 program. From there, teams will go through what is billed as a “comprehensive” agenda aimed at turning their groundbreaking ideas into impactful solutions. It all culminates in December at AWS re:Invent. That said, it’s unclear at this time if any of these startups will participate in the second-annual “Unicorn Tank” competition.

Featured Image: The Amazon Web Services logo hangs atop the expo hall during the company's 2024 re:Invent conference in Las Vegas, Nevada. Credit: Ken Yeung