> ## Documentation Index > Fetch the complete documentation index at: https://docs.spinnable.ai/llms.txt > Use this file to discover all available pages before exploring further. # Worker automation settings > Control how your workers behave in the background — toggle Heartbeat check-ins and Memory Curation on or off per worker. ## Overview Every worker runs two background automation processes by default: **Heartbeat** and **Memory Curation**. These run quietly in the background to keep your workers healthy and their knowledge current. As a manager, you can now toggle each setting independently per worker — giving you precise control over what each worker does on its own, without you asking. These settings are available on the worker's **Settings** tab. You need to be the worker's manager (or a co-manager) to change them. ## Heartbeat **Heartbeat** is a periodic check-in that your worker performs on its own initiative. Think of it as the worker "reporting for duty" — it checks in at a regular cadence, reviews any pending context, and proactively flags things that need your attention. ### What it does * Sends you a brief, scheduled update or prompt based on what the worker has been doing * Reviews outstanding tasks, reminders, and goals * Surfaces anything time-sensitive the worker thinks you should know about * Keeps the worker "warm" even during quiet periods with no incoming messages ### When to turn it off Heartbeat is on by default, and most workers benefit from it. Turn it off if: * The worker has a narrow, reactive role (e.g., it only responds when you message it, and proactive check-ins feel like noise) * You're managing token usage closely and don't need background activity from this worker * The worker is temporarily paused or on standby For workers that handle time-sensitive workflows — like scheduling, follow-ups, or monitoring — keeping Heartbeat on ensures they never miss a beat, even if you forget to check in with them. ## Memory Curation **Memory Curation** is the process by which a worker periodically reviews and refines what it has stored in long-term memory. Over time, workers accumulate facts, preferences, and context from your conversations. Memory Curation keeps that knowledge tidy and accurate. ### What it does * Reviews stored memories for relevance and accuracy * Consolidates duplicate or overlapping information * Flags or removes outdated entries (e.g., preferences that have since changed) * Organizes knowledge so the worker can retrieve it more reliably ### When to turn it off Memory Curation is on by default and rarely needs to be disabled. Turn it off if: * You want to freeze the worker's current memory state exactly as-is (e.g., during a critical project phase where you don't want any background changes) * You're reviewing the worker's knowledge base manually and don't want automated changes running in parallel Turning off Memory Curation doesn't stop the worker from *using* its memory — it just pauses the automatic cleanup and consolidation process. New memories from your conversations will still be added normally. ## How to change these settings Navigate to your worker in the Spinnable app and click the **Settings** tab. Scroll to the **Automation** section. You'll see toggles for **Heartbeat** and **Memory Curation**. Click a toggle to enable or disable it. The change takes effect immediately — no save button needed. ## Choosing the right settings for each worker Not every worker needs the same automation profile. Here's a quick reference: | Worker type | Heartbeat | Memory Curation | | - | - | - | | Executive assistant | ✅ On | ✅ On | | Reactive support worker | ✅ On or ⚪ Off | ✅ On | | Scheduled/monitoring worker | ✅ On | ✅ On | | Standby / rarely used | ⚪ Off | ✅ On | | Frozen knowledge state needed | ✅ On | ⚪ Off | ## Related Understand how memory is built up through conversation and long-term storage Manage the documents and reference materials your worker can search Best practices for keeping a team of workers running smoothly Control background activity to manage your token footprint This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.