Dynamic identity / generative identity
An identity with fixed rules but changing outputs, so the mark can react to context or data without stopping looking like the same brand.
See it
What it is
A dynamic identity keeps a recognizable set of rules while allowing the mark, color, pattern, or composition to change. The fixed part might be a silhouette or grid; the variable part might respond to location, time, content, or supplied data. The 1981 MTV mark was an early flexible identity: its basic shape stayed put while its colors and surface treatments kept changing. Stefan Sagmeister's 2007 identity for Casa da Musica went further, turning views of the Rem Koolhaas building itself into a family of related marks.
Reach for one when variety is part of the subject, as with a cultural institution, festival, media brand, or platform serving many communities. The system should make hundreds of outputs possible without requiring hundreds of separate art-direction decisions.
Gotcha: random is not generative. If every output can change in every way, recognition disappears and the identity becomes a screensaver. Lock at least one strong feature, limit the variable ranges, seed any randomness for reproducibility, and define a static fallback for favicons, invoices, embroidery, and legal documents.
Ask AI for it
Build a dynamic identity for [BRAND] driven by [DATA SOURCE]. Keep [FIXED FEATURE] unchanged in every version, then map one data field to color, one to shape, and one to density using named minimum and maximum values. Use a seeded PRNG so the same input always returns the same output, expose the palette as CSS custom properties, and place every mark in the same SVG viewBox. Generate 12 varied samples plus a static fallback, a one-color version, and a 24px test. Include a short rule table showing exactly what may change and what may not.