July's release centered on deeper Cosmo authoring support, an easier path to building your own AI Interviewer, and a refreshed candidate experience right after an assessment. Here’s what shipped.
Highlights
- Create custom assessments with Cosmo. Cosmo can now author complete custom assessments end to end, not just individual questions. Cosmo cannot yet generate new questions during assessment creation, so look for that in an upcoming release.
- A guided flow for building your own AI Interviewer. Creating an AI Interviewer is now much simpler thanks to a dedicated guided setup flow.
- A refreshed post-assessment experience. The screen candidates see right after finishing an assessment, along with the feedback survey that follows, has been redesigned to match our latest look and feel.
- More control over what syncs to Greenhouse. You can now choose exactly which assessment result fields sync back to Greenhouse, including the option to withhold the score entirely.
Assessments & Hiring
- You can now choose exactly which assessment result fields sync back to Greenhouse and can withhold the score if you prefer not to share it automatically.
- The post-assessment screen and the feedback survey that follows have been redesigned to match our latest look and feel.
- The interview template experience has been redesigned for a cleaner, more consistent setup.
- Waiting room and video call behavior is easier to find and can be configured both at the interview, template, or global level.
- Global interview settings have moved out of a pop-up and into a dedicated page, where you can also configure Fraud Prevention and Acknowledgements account-wide.
- Candidates can only reshare a previous result when the new request uses the same AI-assistance setting as the original, so comparisons stay apples to apples.
Questions, IDE & Simulation
- Frameworks are now queryable through our GraphQL API, including their descriptions, so you can pull framework details programmatically.
- Question "Solution" pages have a new design for both conversation simulation and workspace questions.
- Our GraphQL API now supports creating and editing scorable filesystem questions, including progressive ones.
- You can now query conversation and writing answers through the GraphQL API.
- Execution time limits for free-coding questions are now customizable, which helps avoid timeouts for languages such as Kotlin and C++.
- Rubric configuration for AI-powered grading is now easier to set up.
- Replay is now turned off for a small set of bespoke simulation questions to keep results consistent.
Cosmo & Authoring
- Cosmo can now author complete custom assessments end to end, not just individual questions. Question generation during assessment creation is coming in a later release.
- Cosmo can now author single-function questions too, both progressive and non-progressive, matching this month's Single Questions redesign.
AI Interviewer
- A new guided flow makes it much easier to create an AI Interviewer.
- Live interviews can now include AI-graded questions.