Not long ago, a cover letter did a simple job: It proved a candidate cared enough to spend time. It also offered a quick gut-check on communication, and if the applicant referenced your product or customers, you could squint and call it “fit.”
Then generative AI showed up.
Today, a polished, tailored cover letter can be produced in minutes, sometimes seconds. The writing is clean. The tone is confident. The keywords match the job description perfectly, and it’s often meaningless.
AI hasn’t “broken” hiring. It’s exposed how fragile the cover-letter signal was all along.
What cover letters used to signal, and why that signal collapsed
Cover letters were historically expected to signal three things:
- Effort and motivation (did the candidate care enough to write one?)
- Communication skills (can they explain ideas clearly?)
- Company fit (did they understand the role and your business?)
Even pre-AI, those signals were noisy. Candidates reused templates. Friends edited drafts. Professional services wrote them. There was always a quiet market for “help” that made the letter look better than the person behind it.
Generative AI simply made that weakness obvious at scale. When everyone can produce a crisp, tailored narrative on demand, the cover letter no longer reflects effort, communication, or genuine interest. It’s just formatting. And in high-volume roles, especially engineering, many aren’t read solely by humans anyway, because screening happens upstream through automation and triage.
So small businesses end up in the worst possible place: candidates waste time generating letters, and hiring teams waste time skimming writing that no longer correlates with job performance.
The real risk for small businesses: reacting the wrong way
When SMBs realize applications have been “AI-polished,” they often swing to extremes:
- Over-indexing on polish: assuming the best-written application equals the best candidate.
- Over-correcting with blunt filters: years of experience minimums, pedigree screens, keyword traps—anything that reduces volume.
Both are costly.
Over-indexing on polish rewards candidates who are best at presentation, not execution. Over-correcting can eliminate exactly the kind of talent small businesses often rely on: non-traditional builders, self-taught engineers, and early-career candidates with strong hands-on ability.
The better approach is calmer and more practical: stop trying to outsmart AI-generated narratives and start collecting skill evidence.
What replaces the cover letter as the fastest reliable signal?
If you’re running a small team, you need hiring signals that are fast, reliable, and scalable and that map to the work you need done next week, not the story someone can tell this week.
In practice, three signals beat cover letters for most tech roles:
1) A short, role-specific work sample (30–60 minutes)
Nothing predicts job performance better than watching someone do a slice of the job, under constraints that resemble the job. Work samples and simulations are well-established selection methods because they test applied capability rather than self-description.
For software engineers, that doesn’t have to mean a massive take-home project. It can be:
- A debugging task
- A code review on messy or AI-generated code
- A small feature implemented with clear requirements
- A “fix the failing tests” scenario
- A security-focused exercise that checks judgment and tradeoffs
The key is that it’s job-relevant and scoreable with a rubric. And yes, candidates will use AI. Let them.
The modern question isn’t “Did you write every line yourself?” It’s: Can you produce a correct, secure, maintainable solution and explain why you made the choices you made?
When you measure correctness, reasoning, and judgment, it’s much harder to “fake” competence with a few prompts.
2) A structured screening conversation (often first round)
Unstructured interviews are where bias and inconsistency thrive. A structured interview, with consistent questions and scoring, improves signal quality and the candidate experience.
Historically, the reason interviews happened after short-listing was simple: human time is expensive. But now, scheduling and first-round screening can be made dramatically more efficient.
For SMBs with limited bandwidth, a structured first-round screen can be run:
- By a hiring manager using a rubric
- By a trained recruiter following the same rubric
- Or, increasingly, with AI-assisted tools that standardize questions and scoring (with human oversight where it matters)
The goal isn’t to remove humans. It’s reserving human attention for the moments where judgment matters most, like deep dives on tradeoffs, ownership, and team fit.
3) Proof of work (real evidence, not claims)
Resumes and cover letters are narratives. Proof of work is evidence.
For engineering roles, look for:
- GitHub repos with clear ownership
- Shipped projects (even small ones)
- Meaningful contributions to real software
- Artifacts that show judgment: design docs, tradeoff write-ups, postmortems
For a small business, this matters because you’re rarely hiring for “potential someday.” You’re hiring for execution in a small team where the work shows up immediately.
Skills-based hiring isn’t just a trend. It’s a survival strategy
Across the market, more organizations are moving beyond resumes and adding job simulations, portfolio reviews, and structured methods to assess real-world capability.
For SMBs, the payoff is even bigger:
- Speed: You quickly identify baseline competence
- Signal quality: You measure job-relevant performance
- Fairness: You reduce reliance on proxies like pedigree and polish
- Candidate experience: Strong candidates prefer a process that lets them demonstrate ability
And it aligns with a reality many small businesses already know—hiring the wrong engineer isn’t just expensive; it can delay product delivery, create security risk, and drain the team.
What “good engineering” looks like in 2026
AI isn’t just changing hiring inputs. It’s changing the job. In many teams, engineers are doing less “blank page coding” and more:
- Turning messy business ideas into precise requirements
- Reviewing AI-generated output for correctness and security
- Stitching components together across a real system
- Making tradeoffs under cost, performance, and reliability constraints
A few years ago, a mid-level engineer might spend a week grinding out a thousand lines to ship a couple of features. Today, that same engineer often starts the week wrestling ambiguity into a spec. AI produces the boilerplate quickly, but the real work becomes intent auditing: hunting for subtle logic flaws, bad assumptions, hallucinated dependencies, and security issues that “look fine” until production.
In other words, the role that survives and thrives isn’t “fast typist.” It’s systems thinker + domain translator + reviewer with judgment.
Small businesses should hire for that reality now, because lean teams can’t afford engineers who perform well in interviews but struggle when the work becomes ambiguous.
A practical SMB hiring loop for the AI era
If I were designing a small business hiring process today, with applicants using AI for resumes and cover letters, my guiding principle would be: Ignore unreliable signals and optimize for skill evidence and consistency.
Here’s a practical flow:
- Write a job post that’s specific about outcomes (not laundry lists).
- Invite applicants to a short work sample early (30–60 minutes).
- Use structured scoring to rank candidates (consistent rubric).
- Run a structured first-round screen focused on reasoning and tradeoffs.
- Use humans late in the funnel for deep judgment: ownership, collaboration, domain fit.
- Close the loop by tracking which signals predicted performance and refine the rubric.
The biggest change is psychological. Stop treating hiring as a narrative contest. Treat it as an evidence-gathering process.
The cover letter isn’t “broken.” It’s obsolete.
AI didn’t ruin hiring. It ruined the illusion that cover letters were a meaningful proxy for intent and capability.
Small businesses have an advantage here: you can move faster than big companies, design job-relevant assessments, and build a hiring loop that rewards real skill, not surface-level polish.
In the AI era, the winners won’t be the teams that read better cover letters. They’ll be the teams that measure better signals.
Vikas Aditya is the CEO of HackerEarth, an AI-native talent intelligence platform that helps enterprises hire and grow world-class engineering teams. A veteran of Intel and multiple tech ventures, he has spent more than two decades building products and businesses at the intersection of cloud, data, and developer ecosystems, and holds several U.S. patents in networking and software delivery. He also founded the nonprofit Preserve Culture and is a longtime mentor for student innovation programs.
