Ataraxis: Clinical AI to make disease predictable
Ataraxis builds clinical AI for cancer care. Its models read the pathology slides already made during diagnosis, finding patterns in tumor structure and surrounding tissue that are difficult to measure by eye. Its first product, Ataraxis Breast, uses that reading to estimate a patient's risk of recurrence and the benefit they can expect from treatment. The thesis is that much of what oncologists need to personalize care is already in the tissue, if it can be read at the right scale.
The brand had to earn the trust of clinicians while avoiding two easy registers: the cold abstraction of AI and the soft imagery of wellness. HEX built the identity around the moment the product exists for, when tissue becomes a measurement.
- Services
- Brand Identity: Strategy, Art Direction, Visual Identity, Illustration
Website: Web Design, Webflow Development, Custom Code Animation - Website
- ataraxis.ai
- Year
- 2026
- Raised
- $20.4M Series A
Reading the slide
Our early explorations split along a familiar line. One treated Ataraxis as a data company, with networks of points clustering into the outline of a cell. Another grew organic forms from thousands of small dots, close to tumor or vascular growth. Each captured half of the product.
The final system layers them. Point fields gather into branching, cellular structures, and a technical layer traces paths through them and labels clusters with values between 0.00 and 1.00, the scale on which Ataraxis reports its results. Biology sits underneath, interpretation on top.
Grown, not drawn
We built a procedural tool that grows each structure from a field of points and then annotates it. It rarely produces the same specimen twice, a fitting property for a company whose purpose is treating each patient as an individual case. Ataraxis now uses the same tool, so new imagery for papers, decks, and announcements follows the same logic of tissue and reading.
The register of evidence
Oncologists adopt a new test on the strength of its evidence, so the identity takes its manners from the scientific paper. Statements are set in Reckless, a serif with the composure of a journal, while mono labels carry figure numbers, references, and data the way captions do in print.
On the website, the tests that make up Ataraxis Breast sit on a single product page, each one a different question asked of the same slide. The illustrations run as live code rather than video, staying sharp at every size and redrawing for light and dark mode.
Credits
HEX: Ayush, Maz, Adiel, Xchyler, Emmi, Ivan, Shais
Ataraxis: Jan, Wendy
More work
Plasticity Designed for the power user
Rebrand · 2025
Conversion A modular, multi-product system
Rebrand · 2025 · $28M Series A