Sourceful: UK AI Company Behind the Riverflow Image Models (2026)

Sourceful built Riverflow, the #1 ranked AI image model on Artificial Analysis in 2026. Manchester startup, $31.8M raised, CEO Wing Yung Chan. API via OpenRouter and Runware.

Sourceful is a Manchester-based AI company founded in 2020 by Wing Yung Chan, with $31.8M raised from Index Ventures. Its Riverflow 2.0 models ranked #1 on Artificial Analysis for both text-to-image and image editing, surpassing GPT Image 1.5 and FLUX.2 [max].

Sourceful is a Manchester, UK AI startup founded in 2020 by Wing Yung Chan. The company develops the Riverflow image generation models, which ranked #1 on Artificial Analysis for text-to-image and image editing in 2026. Sourceful raised $31.8M from Index Ventures and partners.

Founded: 2020 · HQ: Manchester, UK · Team: 51-200 · CEO: Wing Yung Chan · Funding: $31.8M (seed led by Index Ventures + Series A, March 2022)

About Sourceful

Sourceful is an AI company founded in 2020 by Wing Yung Chan and co-founders in Manchester, UK. The company started as an AI-powered platform to help brands design packaging and marketing collateral by chatting with an AI, reducing the time and cost of producing brand-consistent physical materials. Chan previously served as CMO at 25 and CTO at 28 at The Hut Group before leaving to build Sourceful. Sourceful raised $31.8 million in total across two rounds. The seed round was led by Index Ventures, with participation from Eka Ventures, Venrex, and Dylan Field, the founder of Figma. A Series A followed in March 2022. The funding has been used to build out Sourceful's core AI research team and develop proprietary image generation models that go beyond what commercially available diffusion models could offer at the time. The company's core research ambition is to push image generation quality specifically for commercial and brand use cases: precise text rendering in images, consistent object placement, color-accurate product mockups, and reliable instruction following across complex visual edits. These are problems that general-purpose image models historically underperform on compared to pure realism or artistic generation. In October 2025, Sourceful released Riverflow 1, a purpose-built image editing model focused on branded packaging and marketing workflows. Riverflow 1 immediately topped the Artificial Analysis image editing leaderboard, becoming the first model from a non-major AI lab to reach the top position. The model used Sourceful's reasoning pipeline combined with open-weights diffusion models to achieve reliable instruction following. In early 2026, Sourceful launched Riverflow 2.0 as a full suite, with three variants: Fast, Standard Preview, Max Preview, and Pro. The Pro tier uses an agentic, multi-step self-correction process to verify prompt fidelity before returning a final image. Riverflow 2.0 from Sourceful ranked #1 in both Text to Image and Image Editing on the Artificial Analysis All Listings leaderboards, surpassing GPT Image 1.5 (high), Nano Banana Pro (Gemini 3 Pro Image), and FLUX.2 [max]. On June 18, 2026, Sourceful announced Riverflow 2.5, adding 4K output, font control, custom scoring, and further extending its lead on Design Arena. Riverflow 2.5 Pro ranked #1 across all three Design Arena categories simultaneously: Image (Elo 1408), Graphic Design (Elo 1469), and Image Editing (Elo 1377). This was the first time any model topped all three Design Arena categories at once. Sourceful distributes its models through API aggregators rather than building its own direct API infrastructure. Riverflow models are accessible on OpenRouter, Runware, Replicate, and WaveSpeed AI. Enterprise and direct API access is available at riverflow.ai. The company does not open-source its model weights. Sourceful's commercial focus sets it apart from general-purpose image labs. The company targets creative agencies, in-house brand teams, and e-commerce businesses that need reliable, repeatable image generation for product photography, packaging mockups, and ad creatives at scale.

Mission

To bring design-grade AI precision to brand creation, giving designers and developers reliable control over packaging, product imagery, and marketing assets at scale.

Products

Links

Website · LinkedIn · Blog · Docs