Established in 2016, Graphcore is a technology start-up based in Bristol, UK, and Palo Alto, CA, focused on machine learning — which, for your sake and mine, I won’t even bother to try to poorly explain in half-a-sentence — and artificial intelligence. They have developed a piece of hardware, a processor, called Intelligence Processing Unit (IPU) designed specifically for machine intelligence workloads where the IPU’s “highly parallel computational resources together with graph software tools and libraries, allows researchers to explore machine intelligence across a much broader front than current solutions” because of course it does. Graphcore’s new identity has been designed by London-based Pentagram partners Jody Hudson-Powell and Luke Powell.
A driver of [Pentagram’s] approach was to strike out against some of the negative connotations of artificial intelligence, including the implicit gender bias that machine learning can inherit from its creators and the data they use. This bias spills into tech branding, which often leans into a world of well-worn masculine tropes and overly complicated language. To challenge this cliché, the Graphcore brand combines hard type and patterns with a soft colour palette and delicate illustrations. In addition, the brand utilises an optimistic and conversational tone of voice that embodies the complexity of advanced technology without becoming too opaque to understand.
Graphcore’s wordmark is drawn from the geometry of the brand’s headline typeface Graphcore Quantized designed by Pentagram. Based on Caslon’s Egyptian, it contains more than 65 alternate characters of varying resolutions. Using opentype features, the typeface seamlessly switches between characters as it’s typed, giving a different yet consistent outcome every time it’s used.
The old logo looked like it belonged to a 1970s computer start-up business, with its extra tight letterspacing and Avant Garde-sque font, which actually looked quite nice as a piece of typesetting but said nothing about what the company does. The new wordmark is a slick mix of bitmap and smooth letters, along with an in-between state, that — if you wanted to get metaphorical — convey the relationship between human and machine and that — if you wanted to get literal — convey some kind of higher-end, computer-based nerdery. It almost seems like a basic solution — computer = bitmap — but there is a particularly nice balance in the letterforms and something disruptively cool about how they can switch around… a feature cleverly implemented throughout the headlines of Graphcore’s website. Twitchy.
The icon… Maybe I’m missing something obvious but I don’t understand why it’s “OOC” unless it’s the emoji for horseshoe mustache : C in which case, hell yeah. Also, sure, it’s a nice-looking icon.
Graphcore’s identity is built around the concept of resolution, with each of its components finding its form by sensing and responding to its environment. To accompany the logo, a visual toolkit of patterns, glyphs, shapes and letterforms was developed that all move and respond to changing resolutions and densities. Aware that the capabilities of machine learning is developing incredibly quickly, the toolkit references design created by neural networks rather than relying on it.
The shapes and glyphs serve as illustrative tools within the system that animate and elevate Graphcore’s content. Finding the right imagery, especially in such a speculative industry, can be difficult, resulting in the use of stock photographs that dilute content or fail to illustrate an idea properly. To combat this, Pentagram developed a shape generator which allows Graphcore’s internal team to create infinite patterns that illustrate their website content, presentations and more. The generator is part-random and part-weighted and is similar to the system developed for Graphcore’s animations, which are used across digital touchpoints. These specially developed tools allow Graphcore to own their brand, an attitude that has been applied to all aspects of the identity, including the introduction of editable digital guidelines and the composition of a Graphcore music library.
It would be almost irresponsible for an identity about machine learning and automated processes for it to not have some kind of flexibility or, even better, its own generator. Although that sounds flippant, this particular generator is quite nice, building on the key aesthetics of the wordmark to generate rather pleasant compositions in a lovely color palette.
The applications manage to strike a great and difficult balance of looking techie but not obtuse, all while building on a fairly standard notion of computers (the bitmap font). Overall, it’s a smart and engaging identity.