All tools
Real job ads · No sign-up · Nothing stored

Automation Scan · Automatisierungs-Check

Agents read a company's own job ads and pull out the tasks people are hired to do, each with the sentence it came from. You correct the list. Then every task is sorted: what a fixed rule can do, what a model can draft for a person to check, and what stays with people.

Company website

The reader follows the careers link and takes job ads from the sitemap the company publishes. About one company in three lists its ads that way.

Or paste a job ad

The text of one ad, or a few lines on what a team does all week. German or English.

0 / 6000
No ad at hand? Try an invented sample:
Model calls: 2 per run · robots.txt honoured · nothing you enter is stored
Agents
Check the list (remove what is wrong, mark what already runs by itself, add what is missing)

Start here

The automation is rarely the hard part. The hard part is the rules around each task: who signs off, what the model may see, and what happens when it is wrong. That is the work I help companies with.

Work with me
Tasks are pulled from the text and sorted by a language model; every task shows the sentence it came from, so you can check it. The flags are prompts for a conversation, not legal advice. Only public job ads are read, robots.txt is honoured, and nothing you enter is stored.