Ammaar Reshi leads product and design for Google AI Studio, Google DeepMind’s platform for building applications with Gemini. He is developing tools that turn natural-language instructions into working software, bringing together interface design, image generation, and the backend infrastructure applications need beyond an initial prototype.
Reshi studied at King’s College London and co-founded HackCampus, an initiative connecting university technology communities with industry. At Brex, he advanced from senior product designer to design director between 2020 and 2024, working on enterprise spending products including Brex Empower, budgets, reimbursements, and bill payments. He subsequently became head of design at ElevenLabs, where he built the design function, redesigned its platform, and helped create its first iOS application.
In December 2022, Reshi used ChatGPT and Midjourney to create Alice and Sparkle, a children’s book about a girl and an AI companion. The project became an early flashpoint in debates over AI-generated art: illustrators challenged the use of their work as training data and questioned its implications for consent, compensation, and employment. Reshi acknowledged those concerns and called for stronger protections for artists. He later experimented with AI-assisted animation, extending his interest in generated imagery into filmmaking.
At Google DeepMind, Reshi and product manager Kat Kampf helped introduce prompt-based application creation in AI Studio. By March 2026, their work included full-stack vibe coding, with Firebase-backed features, multiplayer applications, external libraries, secure integrations, and the Antigravity coding agent. Subsequent updates added Android development, Google Workspace integrations, mobile creation, and export into professional development environments.
- Infrastructure inferred from intent. Reshi wants builders to describe the application they need while AI Studio handles supporting frameworks, storage, integrations, and shared state. His multiplayer racing-game demonstration illustrated why generated software needs working backend behavior, not only an attractive interface.
- Design judgment inside the generation loop. His approach emphasizes typography, visual hierarchy, motion, and deliberate art direction over the repetitive aesthetics common in generated websites. He also uses functional AI-generated prototypes to explore product decisions.
- Controllable multimodal storytelling. With Nano Banana Pro and Gemini, Reshi has explored interactive comics that combine consistent characters, readable embedded text, editable images, and branching narratives. His AI Engineer appearance with Kat Kampf demonstrated how those capabilities can become complete interactive experiences.
