Every other app here sorts jobs the same way, by the federal occupation codes. This one opens that system up: what is in it, where each job is, and what it asks of the people who do it.
The whole SOC tree of US occupations as a treemap, any job or group on a state map, small multiples of every major group, a lookup for 58,000 job titles, and O*NET and Europe's ESCO laid out together by meaning.
From O*NET 31.0, BLS state employment and wages and employment projections, ESCO, and Qwen3 embeddings · sources
Payrolls, postings and announced cuts, set against model releases and everything else that happened in the same years.
US software and tech employment and job postings since 2019, with every major model release marked, and the rate hikes, tax change and layoff wave that came first marked just as plainly.
From BLS payroll data via FRED and Indeed Hiring Lab's job postings index · sources
Every open role at OpenAI, Anthropic, xAI, Mistral and Thinking Machines, read from their public job boards each day: which functions, which skills, where, and the posted pay ranges. The history grows one snapshot at a time.
From the labs' Greenhouse and Ashby job board feeds, saved daily · sources
Every open US data science posting at 80 large employers, from banks and insurers to retailers, drugmakers, tech and the AI labs, read weekly: the skills each company asks for, the titles and seniority it uses, its posted pay, and how the full descriptions group by meaning.
From the companies' public job feeds, O*NET and ESCO skills, and Qwen3 embeddings · sources
Announced US job cuts month by month since 2023, and the share employers attributed to artificial intelligence, against every other reason they gave.
From Challenger, Gray & Christmas's monthly job cuts reports · sources
A job is a bundle of tasks. Which of them could a model do, which do people actually hand to one, and how close do models come to an expert's work?
Pick an occupation and see its tasks, how exposed researchers predicted each one was to language models, how often people really use Claude for it, and what happened to the job's employment and pay.
From "GPTs are GPTs" (Eloundou et al.), the Anthropic Economic Index, O*NET and BLS occupational employment data · sources
How often expert graders preferred a model's deliverable to a professional's on GDPval's tasks from 44 occupations, model release by model release, and what those tasks leave out.
From OpenAI's GDPval paper and dataset, and Artificial Analysis's GDPval-AA runs · sources
If AI takes the routine tasks first, the first jobs feel it first. Is that showing up yet?
Unemployment for recent college graduates against all workers since 1990, by major, and what the payroll studies say about young workers in the jobs most exposed to AI.
From the New York Fed's recent graduates data, BLS, Brynjolfsson, Chandar and Chen's "Canaries in the Coal Mine?" and the studies that answer it · sources