Should your dev CV name the AI tools you use?

Did Replit's switch to OpenAI just make your CV keywords obsolete?

Not obsolete — but it's a clean demonstration of how fast a vendor name on your CV can lose its meaning. On stage at TechCrunch Disrupt on 16 October 2026, Replit co-founder and CEO Amjad Masad said the company was "no longer working with Anthropic." A day later came the clarification: OpenAI is now Replit's default model provider, while Claude stays available to customers who choose it. Masad framed the move as a decision about cost, partnership and performance, and said Replit had been a top-five Anthropic customer until recently. Reporting after the talk suggested the relationship was less severed than the stage comment implied. Either way, the default moved — and most users keep the default.

Here's why that lands on your CV. Plenty of engineers now carry a line like "built and shipped on Replit using Claude," or a skills row naming one model family next to React and Postgres. That sentence described reality in September. After a default switch, it describes a configuration the platform no longer leads with, and a recruiter skimming for eight seconds can't tell whether you picked that model on purpose or simply typed into whichever box was in front of you. The engineering you actually did hasn't changed at all. The signal has — and it now hints at a document nobody has opened since last quarter.

None of this is a one-off event you can wait out. Model providers get swapped, runtimes get acquired, and tool names fold into larger platforms; Cloudflare's acquisition of Deno sent engineers hunting for where Deno, Bun and Cloudflare Workers belong in a stack section. Each of those announcements quietly rewrites a handful of nouns on thousands of CVs. You can't forecast which noun goes next, and refreshing your CV after every press release is a poor use of your evenings. What you can control is structure: build the document so a vendor's business decision costs you one line edit, not a rewrite of your experience section.

Should your CV name the AI tool or describe the capability?

Describe the capability, then name the tool — in that order, and only name tools you could defend under interview questioning. Capability language is what survives a vendor switch: "AI-assisted development," "LLM-assisted code review," "context and prompt design for agentic coding," "testing and security review of AI-generated code." Those phrases stay true whether your editor calls a model from OpenAI, Anthropic or something that launches next spring. Tool names are the perishable layer on top. Treat them like you'd treat a version number: specific, useful, and expected to change. The capability is the claim; the tool is the evidence that you've done it somewhere real.

What hiring teams are actually screening for shows up in the job ads themselves. Postings now ask for "fluency with AI coding tools as a core part of the development workflow, paired with strong judgment and full ownership of AI-generated output," or for engineers who can decide when AI tooling introduces new risk. Read that carefully: the tool list is the entry ticket, and the judgment is the job. So pair every AI mention with the human work around it — the review gate you added, the test coverage you demanded before merging generated code, the time you rejected the model's approach and shipped your own.

Here's the contrarian part. In 2026, a bare "AI tools: Copilot, ChatGPT" line on a developer CV is close to worthless, because every candidate in the stack has it and it differentiates nobody. What differentiates you is what happened when the model was confidently wrong. "Introduced a review checklist for AI-generated migrations after a generated query dropped an index in staging" tells a hiring manager more than any tool list, because it proves you own production outcomes rather than keystrokes. If you only have room for one AI sentence in your experience section, spend it on judgment, not on vendors.

Which AI tool names are actually worth keeping on your developer CV?

The ones in the job ad you're applying to — and across the market that means Cursor, Claude Code, GitHub Copilot and Codex, roughly in that order of frequency. A mid-2026 snapshot of about a thousand hiring companies on one engineering job board found 167 of them naming a specific AI coding tool in their postings, with Cursor appearing at 113 companies, Claude Code at 101, GitHub Copilot at 64 and Codex at 38. Windsurf and Replit trailed in the single digits. A separate posting-level analysis found the same cluster at the top, and a meaningful share of those mentions sat in the required column, not the nice-to-have one.

Read those numbers as a hiring signal rather than a popularity contest. Copilot has far and away the biggest installed base in large enterprises, yet it's named less often in ads than Cursor or Claude Code — because the companies writing "fluency with agentic coding tools" into their requirements tend to be the ones building AI products themselves. If you're applying to that end of the market, naming the tool matters and your examples need to be current. If you're applying to a bank or a hospital network, the capability phrasing carries the weight and the brand is almost decoration.

My practical rule: cap your CV at three or four AI tool names, chosen to overlap with the posting, and drop the rest. Platform brands like Replit belong there only when the role actually runs on that platform. And never write model version numbers into your skills section — "GPT-4o, Claude 3.5" dates your CV faster than a Hotmail address, because a reader can price exactly when you last updated it. Write the tool, not the build. If a job ad does name a model generation, mirror its exact wording in that one tailored copy and leave your master document generic.

Where on your CV should AI-assisted development actually go?

Three places, and no more: a plain skills line, one or two experience bullets carrying numbers, and — if you've built something substantial — a single project entry. Keyword matching still rewards dual placement, so name the tool in Skills and prove it once under a specific role. Keep the skills section as simple comma-separated text under a conventional heading; it's the zone parsers harvest most reliably, and a graphic skill chart or five-star rating turns it into an image a parser reads as nothing. One caveat worth knowing: some systems score skills against your whole document, not just that section, so the experience bullet is doing real work too.

The bullet formula that works is boring and effective: action verb, tool or method, purpose, measurable result. Compare "Used GitHub Copilot during development" with "Rolled Copilot and Cursor into a six-engineer backend team's workflow, cutting boilerplate and scaffolding time on new service endpoints by roughly a third while keeping review coverage at 100%." The second version survives scrutiny because every claim has a unit attached. Reach for metrics you can actually source: PR throughput, review turnaround, flaky test count, incident volume after rollout, onboarding time for new joiners. Round honestly, and be ready to explain how you measured it, because a good interviewer will ask.

What to avoid is just as important. Don't bolt a standalone "AI" block onto the top of a backend CV when the role is distributed systems — it reads as chasing a trend and pushes your actual stack below the fold. Skip tool logos, icons and two-column templates, which routinely scramble extraction and strand your skills in an unreadable text box. If you want the full structural breakdown for engineering roles, the guide on passing the first engineering screen covers how to lay out a stack section parsers can read and how to turn commits and incidents into measurable bullets.

How do you check your CV's AI keywords against a specific job ad?

Run your CV and the job description through an analysis together, rather than eyeballing the overlap yourself. HRLens does this directly: upload your CV as PDF, Word or an image, paste in the job description, and the analysis lists the skills the posting asks for that your document doesn't contain. That's exactly the decision this news cycle forces — whether "Claude Code" or "Cursor" earns its line in your skills section for this application, or whether the ad only ever says "AI-assisted development." The free output gives you a score out of 100, five category scores including ATS compatibility, your strengths and a visual layout check.

Two mechanics worth knowing before you start. You can paste a job URL, but LinkedIn and Indeed block fetching, so copy the posting text instead — it takes five seconds and the targeting is only as good as the text it sees. And every analysis generates ATS-friendly rewritten versions of your CV across six templates that mirror the job's own keywords, which is the fastest way to see what your experience looks like with the posting's vocabulary rather than your own. If you want the full fix list, the "why you might be rejected" breakdown, blind spots, hiring probability and a salary estimate, the Full Analysis runs $3.99.

Build the habit around a master document. Keep one capability-first CV where the AI line reads "AI-assisted development, LLM-assisted code review, automated test generation," then swap in the three tool names each posting actually uses before you send. Re-check whenever you move between families of roles — platform engineering to AI product work, for instance — not for every single application. You can run a free CV analysis to see where your ATS compatibility and tech stack scores sit right now. Do that once before your next batch of applications, and a vendor's default switch becomes a one-word edit instead of a problem.

Frequently asked questions