More detail: What's next, after the course
Optional reading for this lesson. Everything you must do or decide is in the lesson.
This page goes with the Week 4 page What's next, after the course (course/workforce/week4/lessons/anw-w4-l13.md). You don't need it to finish. Read it when you want the longer version.
What you have now
An email team that drafts and waits for your yes, Evy grading every day, the Charter Steward's daily check, Kansa's watch lists, your shelves, your rule sheets, Langfuse recording everything, and the standing rule for anything new. In everyday mode all of it keeps running, and your records are marked as everyday work instead of course practice.
While your computer is off
Your agents run on your computer, only while it's awake and the app is open. Your first brief each day catches up: it runs once what only needs the latest run, replays each missed day where every run counts, and skips what's too late, saying so.
The always-on options, in more detail
| Option | Good for | What it can't do | Where your keys live |
|---|---|---|---|
| Cloud routines in Claude Code | Jobs that only read, like a nightly summary of yesterday | Runs from a fresh copy of your workbench on GitHub, at most once an hour, with no stop to ask you first. Sees only what you saved to GitHub: not your records in runs/, your drafts, your customer mail or your key file. Can't read your chats with your Chief. | Not in your workbench copy. How it signs in to connected tools: check the current docs. |
| ChatGPT's scheduled tasks | Jobs that need only the tools you connect in ChatGPT, and what you upload | Can't reach your workbench folder or any file on your computer. Can't use your agents or read your chats with your Chief. | Never in the chat. How ChatGPT signs in to your tools, and which ones a scheduled task can use: check the current docs. |
| An always-on computer | Everything, all the time | The hardest path: a small server that never sleeps, private network access, and every agent, tool and recorder set up again there. | In a private key file on that server, put there by you. |
The course doesn't teach these. Try them when you're ready, one job at a time, starting with a job that only reads. Product features change, so read the product's current docs before you set one up, and believe the docs over this page where they differ.
Your keys
| Where | Where the key lives | Never |
|---|---|---|
| Your computer | Your Langfuse keys: the .env file in your workbench folder, which Git ignores | In a chat, a screenshot, a note, or a file you save to GitHub |
| A cloud routine or a ChatGPT scheduled task | Signed in through the product itself, by you; how: check the current docs | Pasted into the instructions or the chat |
| An always-on computer | A private key file on that server, put there by you | Copied in from a chat |
If a key ever appears in a chat or a reply, treat it as exposed: make a new one in that tool, put it in place yourself, and, for Langfuse, have the Charter Steward file Evy's grade-write ruling again.
Langfuse after the course
- The recorder records only while it's on, and only in the apps and computers it's wired into. Your standing rule wires in anything new: your Chief flags it, Gigawatt wires it in and proves it by reading a test record back, and the Charter Steward records it.
- Evy writes one score per rule, named
evy.<team>.<rule>, while your filed ruling "Evy writes grades to Langfuse: on" matches your project and key. A new project or a new key means filing the ruling again. - The free Langfuse plan kept records for 30 days when the course was written; paid plans keep them longer. Plans change, so check Langfuse's current docs.
- Evy's local copy of what it needs for grading, in the job's folder under
runs/on your computer, is your long-term history. It stays after the 30 days, and holds what grading needed, not every record. Saving a copy doesn't change how long Langfuse keeps anything. - After you switched to everyday mode, new records are marked as everyday work, not course practice.
The daily check after the course
- Evy's time, if you picked one, is when your morning digest starts Evy's check. Your first brief of the day still runs it if it hasn't run, and it never runs twice in a day.
- If your computer was asleep at your time, Claude Code runs one catch-up when it wakes.
- A scheduled run that reaches a step needing your OK stops, saves a short note, and your next brief tells you. Click "Run now" once while you watch, and approve what it asks. A tool that asks for an OK on every use stops every run; ask your Chief to help take that step out of the morning run.
- To turn it off, ask your Chief to have the Charter Steward record "Turn off Evy's daily check". The latest ruling wins. To pause the morning digest itself, ask your Chief to walk you through it in your app.
Where to get help
Check course/workforce/reference/known-issues.md first. If nothing there fixes it, The Professor can help you write one help post for the Slack channel #agent-workforce-hr, or you can post there yourself. Never put a key, a password or a customer's details in a help post.