Tech hiring has not fundamentally changed in two decades. Resumes get screened, candidates get coding tests, multiple interview rounds, someone gets hired, and teams cross their fingers hoping the decision was right.
Utkrusht AI, a tech startup that recently raised seed funding, thinks this entire model is broken, and has built a platform designed to completely replace it.
The core problem is not that companies lack tools. It is that every tool available today measures the wrong thing.
LeetCode-style tests measure whether someone practiced LeetCode. Take-home assignments measure whether someone had time over a weekend. AI interview bots generate scripted questions that candidates learn to script answers for.
None of it tells you how someone actually performs on the job.
Utkrusht’s answer is what it calls a Watch-them-Work assessment, and it claims to be the world’s first platform built entirely around this model.
How it works
Instead of coding tests or take-home assignments, etc. candidates are placed inside a live production environment and given a real on-the-job task to complete in 30-45 minutes.
Some example tasks can be:
-
A Backend candidate gets a broken payment microservice to fix -
A DevOps candidate gets a broken Kubernetes cluster to fix, optimize a docker container, etc. -
A Fullstack candidate gets a slow API to debug and deploy -
An AI engineer candidate needs to improve embeddings in a chatbot -
etc.
The entire session is recorded. Candidates can use any tools they want, including AI. The platform tracks how they approached the problem, what decisions they made, where they got stuck, and critically, how they used AI. Not whether they used it, but how effectively.
Within 48 hours, hiring teams receive a ranked shortlist of the top 5-10 candidates, each with a detailed report scored across technical execution, problem-solving approach, judgment, communication, and AI usage.
The dashboard shows a straightforward hire/no-hire recommendation alongside the full recorded candidate session.
Utkrusht has assessed over 11,000 candidates on from customers across India, Europe, and USA. Companies using the platform report a ~70% reduction in time-to-hire.
Plus, ~90% candidates complete the assessment during working hours – not on weekends, not reluctantly, because a 30-minute real task is a better use of time than a four-hour take-home.
Why now
The timing has everything to do with AI. The best engineering teams today do not hire for coding speed, they hire for judgment, because AI handles increasing amounts of the execution.
Yet most shortlisting tools were built before this shift and have not adapted. A candidate who scores well on an algorithmic challenge but cannot navigate a messy real-world codebase is becoming an expensive mistake.
Arun Bhimani, VP of Technology at InTech Solutions, who evaluated the platform, said he went from 80 applications to a clear hire — someone now leading a core product initiative – by watching candidates work through the same problems his engineers face daily. Harshal Bhakta, Co-founder of Cital, cut his hiring timeline from months to one week and shortlisted three strong candidates, every one of whom could deliver.
The seed funding will be used to expand the platform’s assessment library, which currently covers 300+ technical skills across fullstack, backend, DevOps, SRE, data engineering, and AI engineering roles.
Getting started
Utkrusht is currently offering a free trial with no credit card required and setup in 5 mins.
For teams not actively hiring, the company is also offering free access to sample candidate reports, real assessment outputs that show what a strong engineering hire actually looks like before you need to make one.
See how it works at utkrusht.ai.