Tasks and Scheduling

Althea’s Automated News Briefing is a background process that monitors external sources for relevant developments. It runs independently to identify important updates you might otherwise miss.

How It Works

Althea uses a three-stage process to filter global news volume down to relevant items using your personal context.

1. Context: Your Dossier

Every search uses your Dossier, a structured record of your expertise and current projects. Althea uses specific interests, like “consensus algorithms” within “distributed systems,” to refine search parameters.

2. Retrieval

Twice a week, the system queries standard news APIs and performs timeboxed web searches (last 72 hours) to identify technical blog posts, preprints, and industry announcements. You can modify this in the settings page, clicking the manage notifications link in the footer of the email, or by simply asking Althea on Email or the web.

3. Filtering

We feed raw search results into an LLM inference step with a specific instruction: discard everything that does not specifically matter to this user. It de-duplicates stories and removes low-value content, retaining only significant updates.

Silence by Design

The system minimizes interruptions. If nothing significant has occurred in your field since the last check, Althea takes no action. You receive a notification only when a relevant event requires your attention.

Utility

The goal is to align the agent’s judgment with that of a close colleague.

Cool example: My (Martin) co-founder Nasim sent me a link regarding the acquisition of Manus, an AI company. Moments later, Althea generated a notification with the same link. Both Nasim and Althea operated on a similar model of my interests. Both recognized that for our specific work, this acquisition was a relevant signal rather than general industry noise. This validation confirms that the Dossier successfully captures the context necessary for autonomous filtering.