Autonomous AI Workflows on a Schedule: A Walkthrough of aiFetchly Task Scheduling
- Step 1: Create an AI Message Task
- Step 2: Configure Tool Risk Tiers and Safety Bounds
- Step 3: Define Cron Triggers and Dependencies
- Step 4: Run and Inspect Execution History
Direct Answer: Power users can automate recurring operational workflows by combining natural language prompts with aiFetchly native Task Scheduler. By assigning granular tool risk tiers, establishing strict execution budgets (max tool calls and runtime limits), and attaching cron triggers, users configure persistent background AI agents that safely execute complex research, data synthesis, and file operations without manual oversight.
Most desktop AI assistants operate strictly through interactive chat windows, forcing power users to babysit recurring research, data synthesis, and system monitoring tasks manually. Whether tracking competitor product changes, compiling morning repository pull requests, or transforming messy web extracts, repetitive workflows belong in an automated background pipeline.
The aiFetchly desktop client resolves this limitation by introducing native background task orchestration. This tutorial walks through building a self-executing AI message task governed by strict safety bounds, explicit tool allowlists, and automated cron triggers.
Step 1: Create an AI Message Task
Open aiFetchly on your workstation and navigate to the Scheduler view in the sidebar. Click New Schedule to launch the creation modal, and select the AI Message task type from the dropdown. In the prompt configuration editor, define your agent objective using clear natural language instructions:
“Search the local knowledge repository for new product updates, summarize key customer-facing features, and write the draft report to the project notes folder.”
Unlike browser-based chat tabs that terminate if reloaded, scheduled desktop tasks run as persistent background daemon jobs that execute reliably across work sessions.
Step 2: Configure Tool Risk Tiers and Safety Bounds
Autonomous agent execution requires deterministic governance. In the task configuration panel, aiFetchly categorizes available tools into four distinct color-coded risk tiers:
- Low (Green): Safe read-only operations like file viewing, local web search, and knowledge base retrieval.
- Medium (Yellow): Operations with moderate operational impact such as creating drafts or formatting data.
- High (Red): Modifying actions like writing local files, executing scripts, or making external network calls.
- Blocked: Tools completely disabled for the current workflow to guarantee strict boundaries.
Enable Auto-Approve Tools only for the specific low- and medium-risk capabilities your agent needs. To prevent runaway recursive loops or excessive token expenditure, establish strict execution budgets as outlined in the task scheduling guide:
- Max Tool Calls: Caps the maximum number of tool invocations allowed in a single execution run (default: 10 calls).
- Max Runtime: Defines a hard termination timeout in milliseconds (default: 300,000 ms / 5 minutes).
- Max Continue Calls: Restricts recursive thought loops to protect your token context window.
Step 3: Define Cron Triggers and Dependencies
Under the schedule trigger settings, choose how your workflow activates. Select Cron Expression to establish a recurring schedule (for example, 0 9 * * 1-5 to run every weekday morning at 9:00 AM). Alternatively, select After Task to chain this workflow to run automatically after an upstream data extraction or scraping job finishes.
Step 4: Run and Inspect Execution History
Save your schedule. aiFetchly registers the task with the native desktop scheduler engine. When triggered, the agent executes silently in the background. Open the Execution History tab at any time to review full tool execution traces, token usage, and generated output files in the auditable log.
If you try it, start with one read-only task and a small tool budget.
Write a comment