Remade

Overview

Remade was a San Francisco-based AI startup founded in 2024 by four Cambridge University computer science graduates. Incorporated as Pheat, Inc., the company entered Y Combinator's Summer 2024 batch with a focused pitch: automate video ad workflows for lifestyle brands using generative AI, replacing expensive agency outsourcing with a three-click pipeline. The product later expanded into a broader "AI-native canvas" — Remade Canvas — offering context-aware generation, character consistency, inpainting, 4K upscaling, and real-time collaboration.

Remade's core failure was structural, not operational. The team built a commercial application layer on top of generative media infrastructure that was simultaneously commoditizing beneath them — and their own open-source contributions accelerated that process. By releasing 8 Wan2.1 video LoRAs that accumulated 250,000+ downloads and became some of fal's most popular endpoints, Remade demonstrated their engineering value to the market while undermining the defensibility of their standalone product.

fal, a generative media platform for developers, acquired Remade on November 6, 2025 — approximately 12–15 months after founding — in what was fal's first-ever acquisition. All four co-founders joined fal as employees. The acquisition price was undisclosed, and given the $500K total funding raised, any return to YC was likely modest. The outcome confirmed the thesis: the team's ML engineering velocity was worth more inside a platform company than as a standalone application business.

Founding Story

Remade's four founders arrived at the company through one of the more technically credentialed founding teams in YC's S24 cohort. All four studied Information and Computer Engineering or Computer Science at the University of Cambridge, and each brought a distinct research or commercial thread that shaped the company's direction.

Christos Antonopoulos, who became CEO, held a MEng from Cambridge with research contributions in machine learning for medical diagnosis and natural language interfaces. ^[1] His background sat at the intersection of applied ML and product — a profile suited to translating research capability into commercial positioning.

Rehan Sheikh, the CTO, had focused his Cambridge research specifically on diffusion models — the architecture underlying the image and video generation systems that Remade would later build on. ^[2] His technical depth in the core generative stack was the team's most direct credential for the problem they chose.

Alex Matthews, Technical CPO, brought the clearest commercial track record: prior ventures had contributed to £700K+ in revenue and funding. ^[3] His presence suggested the team was not purely research-oriented — there was at least one founder who had navigated the gap between building and selling.

Blendi Bylygbashi brought the most eclectic background. At Transport for London, he developed deep learning models for CCTV event detection. His research on intracranial hypertension was published in the Lancet Neurology Journal. ^[4] He also built a YouTube channel — BlenDigi — to nearly one million subscribers, generating $200,000 in ad revenue. ^[5] That last credential is notable: Bylygbashi had firsthand experience with the economics of digital content creation and the friction of producing video at scale — precisely the problem Remade set out to solve.

The specific moment or conversation that brought the four together is not documented in public sources. What is clear is that the founding insight was grounded in lived experience: creating video content for social platforms was expensive, slow, and dependent on agency intermediaries that charged thousands of dollars for work that generative AI could, in theory, compress into minutes. Bylygbashi's YouTube background gave the team a credible user-side perspective on that pain.

The company was incorporated as Pheat, Inc. — a holding entity name that suggests the founders anticipated the brand or product might evolve. ^[6] "Remade" was the product identity layered over that legal structure, a pattern common among YC teams that enter the program with a thesis but expect to iterate on the specific form.