Charlie Axelbaum
Activeaisoftware

Nightloop

A product exploring intelligent recommendations and signal-driven interfaces.

Nightloop is a nightlife discovery product built around dynamic signal rather than static listings. Recommendations are shaped by timing, context, crowd, and pattern, with AI used selectively for inference and normalization, not as the core mechanism.

The Problem

Existing nightlife discovery relies on static venue listings, user reviews, and manually curated collections. None of these capture what actually drives a good night: timing, crowd composition, energy at a given moment, and the gap between a venue's average reputation and its state on any specific night.

The Approach

Nightloop is built around a dynamic signal layer rather than a static content layer. Recommendations are shaped by:

  • Time and context (day of week, hour, upcoming events)
  • Crowd and energy signals derived from available data
  • User pattern and preference history
  • Proximity and real-time availability
  • The interface reflects current conditions rather than historical averages.

    AI Layer

    AI is used selectively: for normalization of inconsistent venue and event data across sources, for inference when direct signal is unavailable, and for summarization at the recommendation interface. The goal is useful inference. AI is not the product; the signal layer is.

    Status

    Early product development. The starting vertical is nightlife in a single market. The longer-term question is whether the signal approach generalizes to other discovery problems where real-time context matters more than static reputation.

    Stack

    TypeScriptNext.jsPythonPostgreSQL

    Timeline

    Started

    January 2025