Skip to main content
This example demonstrates Fliiq’s daemon, scheduled jobs, and progressive memory — a daily fitness coaching job that adapts based on your history.

The Setup

A user wanted daily workout reminders tailored to their tennis training goals. They asked Fliiq in a chat session:

What Fliiq Created

The agent called create_job and wrote this YAML:

How It Works

First Run

The agent has no history, so it creates a baseline:

Subsequent Runs

Each day, the agent reads its memory, sees what was done previously, and varies the workout:

Over Weeks

The memory accumulates workout history. The agent tracks progression:

Adding Telegram Delivery

After a week of the job running, the user wanted daily workouts delivered straight to Telegram instead of just logging to the audit trail. They told Fliiq:
Fliiq investigated and found that its own job delivery system only supported email and SMS — Telegram wasn’t wired up yet. In a single conversation, the agent:
  1. Read the job config and found delivery.type: telegram was already set
  2. Diagnosed that the executor’s _build_delivery_hint() function had no Telegram branch
  3. Added Telegram delivery support to the executor
  4. Added send_telegram to the job’s skills list
  5. Sent a test message to verify — delivered instantly
The whole fix took about 60 seconds. The next scheduled run delivered a full workout plan to Telegram:

Other delivery options

SMS delivery works the same way:
Or email:

Managing the Job

The Key Insight

The combination of scheduled execution + per-job memory + delivery channels creates a persistent, adaptive agent that reaches you where you are. Each run builds on the last. The agent isn’t just generating random workouts — it’s tracking your progression, varying exercises, adapting to your performance, and delivering results to your phone. When the delivery channel wasn’t wired up, Fliiq diagnosed and fixed the gap itself — no manual coding required. This self-fixing pattern means Fliiq’s capabilities grow as you use it. This pattern applies to any domain: language learning, daily journaling, project status tracking, email digests with deduplication.