Artificial intelligence is moving beyond chatbots and copilots into the infrastructure, development environments and specialized systems that businesses increasingly rely on to build and deploy AI-powered products. A new generation of founders is tackling different layers of that stack, from autonomous software engineering and accessible app creation to model infrastructure, enterprise data and specialized AI compute.
These five founders are leading companies that reflect some of the most important directions in the current AI market. Their approaches differ, but each is working on a fundamental question: what does the technology stack need to look like when AI becomes a core part of how software is built and operated?
Matan Grinberg — Co-founder & CEO, Factory
Matan Grinberg is the co-founder and CEO of Factory, an AI research and software company focused on bringing greater autonomy to software engineering. Factory’s central product is its “Droids,” autonomous AI systems designed to work across software development tasks including coding, testing, debugging, refactoring, documentation and code review. Grinberg has positioned Factory around a broader goal than simply helping individual developers write code faster: the company wants AI agents to become an interconnected system operating across the software development lifecycle.
That vision has expanded with Factory 2.0, which Grinberg describes as a “software factory” built around agent-native development. The platform is designed to connect signals such as bug reports, customer feedback and business requirements with development workflows, while allowing organizations to use different models depending on the task. Factory has also expanded the ways its Droids can operate, including through interfaces such as the CLI, IDE, Slack, Linear and browser environments.
Amjad Masad — Co-founder & CEO, Replit
Amjad Masad is the co-founder and CEO of Replit, the software development platform built around making software creation accessible beyond traditional programmers. Replit’s vision has increasingly centered on AI, allowing users to describe what they want to build in natural language and then use AI-powered tools to create, launch and grow applications. The company’s broader mission is to reduce the barriers between an idea and working software, making software creation more accessible to people without conventional programming backgrounds.
Masad brings a deep developer-tools background to that mission. Before Replit, he worked at Facebook, where he oversaw JavaScript infrastructure and contributed to widely used open-source developer tools, and he was also a founding engineer at Codecademy. Under his leadership, Replit has increasingly framed AI not simply as another coding assistant, but as a way to change who can participate in software creation and how quickly ideas can become functioning products.
Lin Qiao — Co-founder & CEO, Fireworks AI
Lin Qiao is the co-founder and CEO of Fireworks AI, an AI infrastructure company focused on training and inference for open and specialized models. Qiao previously led PyTorch at Meta, and Fireworks was founded with a team whose backgrounds span PyTorch, Meta and Google. The company is building infrastructure designed to help organizations take powerful open models and turn them into specialized intelligence that can be deployed for specific applications and industries.
Fireworks’ strategy increasingly emphasizes giving companies control over the models and infrastructure underlying their AI applications. Its platform supports model training and inference while also providing access to open models and specialized deployments. Qiao has also become an active voice in the debate over whether companies should simply rent intelligence from major model providers or develop more differentiated AI capabilities around their own data, models and workflows.
Jerry Liu — Co-founder & CEO, LlamaIndex
Jerry Liu is the co-founder and CEO of LlamaIndex, a company focused on helping AI agents work with real-world enterprise data. The project began in 2022 as GPT Index, an effort to help large language models access information outside their training data, before becoming LlamaIndex. Liu’s original thesis was that increasingly capable language models would still need infrastructure to connect them to private databases, documents and other external information.
That idea has become increasingly relevant as enterprises move from experimenting with generative AI toward deploying agents in production. LlamaIndex now focuses heavily on agentic data workflows, with LlamaParse positioned as an agentic OCR platform capable of turning complex documents into structured, agent-ready context. Liu has also been focused on problems such as document extraction, long-horizon agents and the infrastructure required for AI systems to reliably operate on business information.
Andrew Feldman — Co-founder & CEO, Cerebras Systems
Andrew Feldman is the co-founder and CEO of Cerebras Systems, the AI computing company built around wafer-scale processors and high-performance AI infrastructure. Cerebras was founded in 2015 with the goal of developing wafer-scale computing, an ambitious approach that placed an entire processor on a much larger silicon wafer than conventional chips. The company has since built systems designed specifically for AI training and inference and has expanded its offerings to both enterprise deployments and cloud-based access.
Feldman’s background spans multiple generations of computing infrastructure. Before Cerebras, he co-founded and led SeaMicro, which was acquired by AMD, and previously held executive roles at Force10 Networks and Riverstone Networks. Cerebras entered a new phase in 2026 with its public-market debut and the introduction of its CS-4 system, while continuing to focus on fast AI inference and large-scale computing infrastructure.
