At DevDay 2026, Alex Mashrabov introduced Higgsfield’s AI platform that enables creative marketers and developers to build end-to-end, interactive advertising campaigns by autonomously managing AI models and GPU resources for continuous, high-quality media production. Highlighting the platform’s ability to democratize creative workflows and ensure secure, scalable deployment, Mashrabov encouraged developers to leverage evolving AI technologies to create innovative media experiences, sharing that Higgsfield’s tools and video editing skills are now open-sourced to support this mission.
At DevDay 2026, Alex Mashrabov, co-founder and CEO of Higgsfield, introduced the company’s innovative multimedia AI platform designed to empower creative marketers and developers. Higgsfield enables users to build end-to-end advertising campaigns using advanced AI models like Astra, alongside proprietary image and video models. The platform not only helps businesses grow revenue through AI-generated ads but also addresses unmet needs in creating new forms of interactive and creative media. Higgsfield’s API allows developers to ship these novel experiences, enabling even small teams to produce interactive shows that were previously unimaginable.
A key challenge Higgsfield tackled was delivering continuous, high-quality media experiences rather than just isolated assets like concept images. To address this, they developed an intelligent agent capable of selecting appropriate AI models and managing GPU infrastructure autonomously. This agentic system evaluates output quality and makes real-time decisions to ensure consistent execution across various use cases, such as advanced video editing. The ability to spin up GPU resources dynamically was critical to supporting complex workflows and delivering scalable, reliable media production.
Mashrabov highlighted the transformative impact of Astra and similar models on creative workflows, particularly in reducing the need for large teams and multiple specialized tools. Traditionally, producing high-quality visual content required extensive coordination among many professionals, but Astra’s capabilities allow smaller teams to automate and streamline these processes. This democratization of creative production is exemplified by Higgsfield’s work with popular streamer Neon, where they enabled a live, 10-hour AI-generated interactive stream that reconstructed a realistic environment in real time—a feat previously thought impossible.
The conversation also touched on the importance of controllability and security in deploying AI-driven media solutions at scale. Higgsfield’s agent incorporates guardrails and perimeter infrastructure to ensure secure and reliable GPU deployment, which has been essential for their enterprise go-to-market strategy. By making these capabilities available through their API, Higgsfield empowers other developers and businesses to build innovative, interactive media experiences with confidence and scalability.
Looking ahead, Mashrabov encouraged developers to embrace the rapid evolution of AI models and focus on building interactive media experiences that leverage these advancements. He shared that Higgsfield has open-sourced AI video editing skills used internally to create educational content, illustrating a new use case where AI helps generate tutorials and product videos. Concluding with an inspiring message, he emphasized that Higgsfield was built in just 18 months and that now is the best time for developers to start building their own AI-powered creative solutions.
Useful Links
- Higgsfield API Documentation — Essential for developers to understand and use the Higgsfield platform capabilities described in the video.
- Astra AI Model by Higgsfield — Central AI model discussed as a core technology in Higgsfield’s platform.
- Higgsfield Open Source AI Video Editing Skill — Directly related to the new use case and tool introduced in the video for AI-assisted video editing.
- Research on Agentic Systems for AI Model Selection and GPU Provisioning — Supports understanding of the agentic system architecture and decision-making described in the video.