AI graphic design tool Recraft gets $12M to launch a foundation model

London-based Recraft, a startup offering brands an AI-powered graphic design generator, today announced it has raised $12 million in a series A round of funding.

The investment has been co-led by Khosla Ventures and former GitHub CEO Nat Friedman, with participation from RTP Global, Abstract VC, Basis Set Ventures, Elad Gil and other angels.

Founded by Anna Veronika Dorogush, who created CatBoost open-source gradient boosting library, the startup said it plans to use the capital to build its product and train its own graphic design foundation model that could help brands not only generate consistent designs as per their color palette and style but also iterate on them quickly.

It joins a bunch of AI startups vying for the elusive creative AI space – a market expected to grow at a CAGR of over 34% between 2022 and 2032.

What does Recraft bring to the table?

While there are many tools to create AI-based art, most prominently Midjourney, Stable Diffusion, and DALL-E 3, Recraft differentiates by targeting the niche professional segment or brand-centric graphic designing.

The startup offers a web platform, where users can sign up and start giving text prompts for generating not just raster images – illustrations and 3D images – but also scalable vectors like logos and icons that can be exported as SVG files to support professional applications. There are scant similar solutions currently available targeting graphic designers, though research has been published on new models, such as COLE, that would address this sector.

More importantly, in addition to text-based generation, the company also provides users with a bunch of additional tools that help with controlling and modifying their AI rasters and vectors.

This includes basic tools to adjust the size and style of the image (like choosing the geometry of icons, the style of lines, or the type of illustration) as well as complex ones such as customizing it according to a brand’s color palette, complete with different layers of colors as well as the desired background.

Users can even define the level of detail – ranging between primitive and high – and mix one style with another to create something new. 

All this put together allows graphic designers to create consistent brand assets in a matter of minutes. And the impact is already beginning to show. The platform debuted about eight months ago and has already been adopted by over 300,000 users.

However, it remains unclear how many of these are paying customers since the company offers both a free and paid tier priced at $20/month. The questions sent by TechForgePulse remained unanswered at the time of writing.

Goal to build a proprietary foundation model

Now, with the platform up and running and a strong, growing user base in hand, Recraft is looking to take things to the next level. With this funding, the company plans to advance its use of machine learning and train its own foundation model to generate even better style-consistent design elements, sophisticated imagery and graphic designs.

It further plans to add capabilities such as visual referencing with brand imagery for better style control and the production of graphic designs combining text and images, like social media posts.

“Recraft was created to meet the requirements of professional design, generating consistent style images and high-quality vector illustrations suitable for everything from advertising and digital marketing assets to iconography. And the development of our Foundation Model will enable us to improve the quality and consistency of the AI-generated graphic design.” Dorogush said in a statement.

While it remains unclear which model is being used by the company at present, training one from scratch could give the company an edge in the creative AI space, potentially allowing it to draw customers who do not have the resources to set up big design teams. In the longer run, and with significant upgrades, it may even eliminate the need for designers.

According to Precedence Research, the market of design-centric generative AI is expected to grow from $412 million in 2022 to $7.75 billion by 2032. Other known players currently operating in the category are Typeset, Kittl and Jasper. Not to mention, industry leaders like Canva and Adobe, who already have a strong customer base, have also added gen AI-powered capabilities into their respective products.