Due to the competitive nature of the artificial intelligence space and DeepCube's advancements in "sparsified" neural networks, they wanted to create an identity that spoke to their expertise but didn’t expose the details of their approach.
We worked alongside DeepCube's leadership team to abstract their message through specific iconography and language that appeals to potential partners while keeping competitors uninformed about the details of their activities.
What we did
A dynamic brand identity designed to mirror the traits of human personality.
What Is Edge Focused Deep Learning?
Traditionally, artificial intelligence applications utilize centralized cloud platforms, to which devices send along their data for interpretation. Deep learning at the edge means that individual devices have a built-in ability to interpret their data, without a need to connect to the cloud.
Power in a small package
To bring deep learning directly to individual devices, artificial intelligence functionality must utilize either a central processing unit or a graphics processing unit, a requirement that previously limited functionality due to constraints with memory or processing power.
DeepCube created neural network technology that is efficient enough to operate on either a CPU or a GPU and provide powerful and fast deep learning functionality without running into hardware limitations. It is a self-contained module that can easily be plugged into the native code of a range of software platforms.
DeepCube is a brain in a box
DeepCube's modularity and power are most easily visualized as a tangible product. As we conceptualized the identity, the platform was often referred to as "a brain in a box," a quite literal interpretation of "Deep Learning Cube."
DeepCube's elegance is that their name describes both the company and their product. We knew that the mark must also work double duty, communicating brain-like intelligence in a neat little cube-shaped box. The resulting mark is a memorable cubic framework that contains a network of neurons and synapses.
An identity that behaves human
The logo's beauty extends beyond its alluring visual composition. Just as AI resembles human thought, the logo mirrors human personality by tailoring itself to every situation it encounters, consistently presenting that version of itself at every subsequent meeting.
First meeting
The mark examines the individual, event, or situation and presents a unique version of itself tailored to that encounter.
Every subsequent encounter
The mark recognizes the individual, event, or situation, recalls how it presented itself originally, and presents that version of itself. The anatomic basis of every logo is consistent, however, their unique composition is formed through individual experience. Activity stimulates thought signals that fly through the neural framework.
The mark comes to life
Thought signals traverse the avatar mark, connecting nodes and maintaining a constant state of motion that expresses the ceaseless progress of Artificial intelligence. Color is also used to highlight the position of activity, and follow the signals around the mark.
Taken from neural networks
DeepCube operates through an approach called neural network "sparsification," where large complex networks are simplified through a regime of constant training. Our avatar is created by assessing all possible node and layer combinations and keeping only those that visually resemble the initial logo.
A fluid approach to typography
The typography is as varied as the brand mark, mixing multiple weights from thin, to bold, to black. GT America Expanded and Extended offer a wide stance, intentionally contradicting a mark with "deep" in its name to signify that AI doesn't have to adhere to predetermined rules.
Titles and messages move on the vertical axis, rising and sinking into the horizon to extend the notion of depth that is inherent to the brand.
Informative iconography
Iconography is used for both literal and abstract communication of ideas, including the capabilities, use cases, or differentiating factors of DeepCube's platform. Icons always exist within a cube or breaking out of one, so as to seamlessly work with the core mark.
Naturally interactive
More than just an online brochure, the landing page itself is a feat in technical engineering. Multiple layers of WebGL Canvas bring a neural network to life with the help of a particle system that makes efficient use of th GPU, similar to DeepCube's platform.
For investors and partners a more robust version of the website dives deeper into the technical characteristics and advantages of the platform, complete with interactive elements that explain neural network training, inference speeds, and accuracy over time.
The technologies we used for the website:
WebGL
Lindenmayer tree (L-system)
GPU particle system
PostProcessingVertex Buffer Object (VBO)
Frame Buffer Object (FBO)
A brand built with empathy
Created through a study in human psychology and self awareness, we developed an identity that connects the digital and physical worlds, and uses intelligent technology to recognize and respond to individuals and experiences.
The identity speaks to the company's name and the power of their platform
Rainfall’s partnership with DeepCube resulted in a dynamic brand expression that uses deep learning to recognize its audience and respond with a unique logo based on a standard visual framework.
By creating a useful application from powerful new technology, DeepCube's identity is an expression of their core values and an example that science and art can combine to create beautiful results.