Two-Day Work-Test Hire
Bring finalists on-site for two days to do a real end-to-end project — it screens skill and fit at once.
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 4
- Confidence
- 90%
Anysphere's core hiring mechanism is a two-day, in-person work-test where a candidate does a real (but non-shipped) mini project largely end-to-end in the company's codebase. It reveals an actual work product rather than interview performance, and because you spend two days around the person, it doubles as a 'do you want to be around this person' test. Truell expected it not to scale and was surprised by its staying power.
Origin
Michael Truell describing Anysphere/Cursor's evolved interview loop.
Core principles
- 01A real work product beats interview theater for predicting on-the-job performance.
- 02Spreading the equivalent of a half- or one-day onsite over two days gives the candidate room to actually build.
- 03Extended contact is itself the culture-fit test — you're around the person through multiple meals.
- 04Early on, before the product is impressive, the team and the experience of working with it is the main recruiting asset.
How to run it
- 1
Build a canonical project list
Prepare a set of realistic mini-projects, often in your actual codebase, that a candidate can complete end-to-end in two days (work you don't ship/use).
Pro tip It doesn't have to be time-intensive for the team — spread the effort you'd spend on a one-day onsite across the two days.
- 2
Run it in person over two days
Have the candidate on-site doing the project largely on their own, with some collaboration, sitting with the team.
Pro tip Share multiple meals together so the fit signal is real, not staged.
Watch out This leans on being an in-person company; remote versions lose the 'be around this person' signal.
- 3
Evaluate the work product AND the fit
Judge the real end-to-end output and separately ask 'do we want to be around this person' based on two days of contact.
- 4
Use it to recruit, not just screen
Let the candidate experience the team so, especially pre-product-maturity, they get convinced they want to throw in with you.
In the wild
Truell thought the two-day work test 'would not scale past a few people,' but it had 'surprising staying power,' becoming core to both evaluation and to getting early candidates excited when the product itself wasn't yet impressive.
→ A durable, dual-purpose hiring filter that helped Anysphere assemble a world-class 60-person, engineering-heavy team.
Common mistakes
Relying on interview questions instead of real work
Standard interview questions don't show a real work product; the two-day project reveals how someone actually builds end-to-end, which questions can't.
Assuming an intensive work-test can't scale
Truell assumed it wouldn't scale past a few people and nearly under-invested; spreading team time across two days made it scalable and it became their most important evaluation tool.
Is it for you?
Best for
In-person startups hiring engineers/researchers where work-quality and culture-fit both matter enormously.
Not ideal for
Fully remote teams or high-volume hiring where a two-day in-person commitment isn't feasible.
From the transcript
“we have people on site for two days and do do a project with us a work test project”
“this two-day work test, which we thought would not scale past a few people, has been has had surprising staying power”
“it really helps you it helps to enforce you know do you want to be around this person type test um because you are around…”
“here's a real mini two-day project you're going to do it end to end largely being left alone”
From the episode
The rise of Cursor: The $300M ARR AI tool that engineers can’t stop using
Michael Truell (co-founder and CEO)