When you paste a job description to start a blueprint, the same pass that extracts tech skills and question types also runs two other checks on that JD text — you don't need to do anything extra to get them.
The first is a timeline check: it flags any requirement that asks for more years of experience with a named technology than that technology has actually existed for — for example, a JD asking for "10+ years of Kubernetes" experience. Each flag shows the specific requirement and the specific concern.
The second checks the JD for language worth reconsidering across five fixed categories: age-coded language, gender-coded language, disability-unrelated requirements, citizenship or national-origin language, and salary-history requests. Every category is always shown, whether or not it found anything — a clean category just shows as reviewed with nothing to reconsider, so you can see at a glance that all five were actually checked, not skipped.
You'll sometimes also see a tone observation — a softer, informal note about how the JD reads overall. This is deliberately kept separate from the five bias categories above and isn't itself a flag.
You'll see all of this on the review screen right after pasting a JD, and it isn't a one-time thing — it's also available afterward from the role's page, so you can revisit it any time you're editing the JD or blueprint later. Like every other AI check in the product, this never blocks you from saving and never assigns a score — it just surfaces things worth a second look before you post the role.
This check only looks at the JD text itself. For the equivalent check on a candidate's resume, see Resume timeline check & skill extraction.