Understand why continuity and memory matter, and how Whiskers makes it happen on your own device.
AI systems forget previous sessions, context windows are limited, and conversations reset or degrade over time. Local models have no built-in memory, so every interaction starts from scratch.
Whiskers adds a runtime around the model: persistent logs, bounded context rebuilding, commands, and optional summaries for longer-running work.
Typical AI sessions are disposable. They may remember the current browser chat for a while, but the UI is fragile and the model has no durable source of truth once the session resets.
Whiskers records every interaction in an append‑only session log. This persistent history survives restarts and UI reloads, so your conversations never disappear.
Multiple browser tabs or devices can connect to the same conversation. The UI isn’t the source of truth; instead, the log is shared so you can pick up wherever you left off from any session.
Whiskers uses background summarization to compress older content into concise summaries. This keeps the context window lean while preserving long‑term memory for multi‑day conversations and ongoing projects.
Your conversations stay on your machine. Whiskers works with your own AI model running locally, and nothing is sent to a remote server. You retain full ownership of the data and the model.
Whiskers employs two models: a primary model for conversation and a secondary summarization model running in the background. This dual‑model approach keeps interactions smooth while enabling summaries for long‑term memory.
