How AI Changed the Tech Hiring Process, on Both Sides of the Table
AI made applying nearly free, so application volume rose and quality signals collapsed. Companies answered with automated screening, skills tests, referrals, and more live or in-person interviews. For candidates, the practical shift is away from polished documents and toward evidence that is hard to fake: shipped work, referrals, and verifiable history.
The volume problem
A few years ago, a tailored application took an hour. With AI tools it takes a minute, and tools exist that apply to hundreds of roles automatically. Postings for remote developer roles can receive far more applications than any team can read, many of them near-identical in structure and phrasing.
The effect is that the documents that used to carry signal no longer do. A well-written cover letter shows that someone had access to a language model. A CV that mirrors the job description shows that someone pasted the job description into a tool. Hiring teams know this, and they discount accordingly.
What companies did in response
Hiring teams did not stop reading applications. They changed where they look for signal, and moved it to stages that are harder to automate or fake.
Automated screening
Applicant tracking systems increasingly rank or filter candidates with AI. This handles volume, but it filters on patterns in the text, which a candidate using AI to tailor their CV can match. The result is an arms race between generated applications and automated filters, with genuine candidates caught in between.
Skills assessments earlier
More processes put a short practical task or technical conversation near the start, rather than after several rounds of CV review. A real task is harder to fake than a document, although take-home work can also be AI-assisted.
Referrals and communities
Referred candidates come with a human vouching for them, which is exactly the signal a flooded pipeline lacks. Many teams now source first from their network, open-source contributors, and developer communities before opening the public funnel.
Live and in-person rounds
AI interview assistants and identity fraud in remote hiring have led some companies to reintroduce in-person or tightly proctored final rounds, especially for roles with access to sensitive systems.
What changed for candidates
The practical lesson for candidates is that effort spent polishing documents has falling returns, and effort spent on proof has rising ones.
- Apply to fewer roles, with a reason for each. Mass applying competes with automated applicants on their terms. A short, specific note to a hiring manager about why this team and this problem stands out because it cannot be generated at scale.
- Make work inspectable. Public repositories, shipped projects, write-ups of problems solved, and contributions to real codebases give a reviewer something to check.
- Use referrals honestly. Ask people who have actually worked with you. A referral from a stranger who glanced at your profile carries little weight and can backfire.
- Prepare for live evaluation. Expect to explain your own past work in depth, reason through a problem aloud, or pair on code. These formats reward real understanding.
What changed for hiring teams
Teams that hire well in this environment design for signal rather than volume. They write postings specific enough that unsuitable candidates self-select out. They decide what evidence would show a candidate can do the job, and ask for it directly. They use automated screening for logistics, such as location or work authorisation, rather than for judgement. And they verify identity and history before extending access to systems, because the cost of a fraudulent hire has risen along with the ease of committing one. The best of them also tell candidates what the process looks like up front, which improves both the applicant pool and the experience.
Where this is heading
The underlying shift is from claims to evidence. When anyone can produce a convincing claim, the scarce thing is a record someone else can verify: work that was delivered, reviewed, paid for, or attested to by a counterparty. That is why verifiable credentials and onchain records of completed work are drawing attention in technical hiring. They are not a replacement for interviews, but they move part of the trust question from a document the candidate wrote to a record the candidate cannot easily fabricate.
Frequently asked questions
- Is AI screening rejecting good candidates?
- Sometimes. Automated screening ranks applications by patterns in the text, so a strong candidate with an unconventional history or a CV that does not echo the posting's wording can be filtered out, while a tailored generated application passes. Teams that care about this use automation for logistics and keep human review for judgement calls.
- Should I use AI to write my job applications?
- Using it to check clarity and fix errors is reasonable. Using it to mass-produce tailored applications mostly puts you in competition with automated applicants, and reviewers are good at recognising generated text. A short, specific, genuinely personal note to a relevant person usually outperforms many polished generic applications.
- Why are companies bringing back in-person interviews?
- Mainly because of fraud and AI assistance during remote interviews. Some candidates use tools that feed answers in real time, and some remote hires have turned out to be different people from those interviewed. An in-person or carefully proctored round confirms identity and shows how a candidate reasons without assistance.
- What evidence matters most to hiring teams now?
- Evidence that is hard to fake: shipped work you can explain in depth, public contributions reviewed by others, referrals from people who worked with you, and history that can be verified independently. Documents still matter for logistics, but they carry less weight on their own than they used to.