How to Connect OpenAI to Your ATS (Step by Step)

Only 19% of core HR processes in Europe are currently enhanced with generative AI, according to McKinsey's HR Monitor 2025, which means most hiring teams are still screening CVs by hand while a smaller group quietly automates the bottleneck. Connecting OpenAI to your applicant tracking system is one of the fastest, cheapest ways to close that gap, and it takes far less technical effort than most HR teams assume.

Quick Answer

Connecting OpenAI to your ATS means generating an API key from your OpenAI account, entering it into your ATS's AI or integrations settings, choosing a model, and testing it on a live job before rolling it out. Most modern ATS platforms, including Klearskill, support this natively and the whole process takes under 15 minutes.

What You'll Learn in This Guide

• How to generate and secure an OpenAI API key for hiring use
• Where to enter it inside a typical ATS or CV screening tool
• Which OpenAI model to choose for CV screening versus other hiring tasks
• How to test the connection safely before screening real candidates
• What to do if the connection fails or screening results look wrong

Why Connecting OpenAI to Your ATS Matters

Hiring teams that connect their own AI key are, in effect, buying compute at cost rather than paying a vendor markup on every CV screened. This is exactly the model behind Klearskill's Starter plan, which lets teams bring their own OpenAI, Anthropic, Gemini or OpenRouter key and screen candidates for pennies per screen rather than paying a fixed AI-included fee.

The efficiency gains are well documented. LinkedIn's Future of Recruiting report found that recruiters who actively use generative AI tools save approximately 20% of their working week, equivalent to a full day, largely because they are no longer manually reading every CV that lands in a pipeline. Meanwhile, SHRM's Human Capital Benchmarking data shows the average time to fill a role sitting at 42 days, a number that AI-assisted screening can meaningfully compress by removing the manual read-through stage that eats the first week or two of most hiring processes.

There is also a cost angle specific to bringing your own key. Rather than paying a vendor for bundled AI usage you may not fully use, connecting your own OpenAI account means you only pay OpenAI directly for the tokens you actually consume, which for most CV screening workloads amounts to a few cents per candidate.

The competitive pressure to make this switch is building. Gartner predicts that by 2027, as many as 75% of hiring processes will incorporate some form of AI proficiency assessment, and hiring teams that have already automated CV screening will be far better positioned to adapt than those still building the workflow from scratch. Flexibility in how a team hires matters too: CIPD's Resourcing and Talent Planning research found that 54% of organisations facing recruitment difficulties are responding by offering greater flexibility, and faster, AI-assisted screening frees up recruiter time to actually manage those flexible arrangements rather than getting stuck in a CV backlog.

Step 1: Create an OpenAI Account and API Key

Start at platform.openai.com and sign up for an account if you do not already have one, using a work email tied to your hiring team rather than a personal address. Once logged in, navigate to the API keys section under your account settings and generate a new secret key.

Name the key something identifiable, such as "ATS-Screening-Production," so you can track its usage separately from any other OpenAI projects your company runs. Copy the key immediately, since OpenAI only displays the full string once. If you manage multiple hiring tools or multiple recruiters who might need visibility into usage, consider creating a separate OpenAI project first, then generating the key inside that project so its costs and logs stay isolated from unrelated company usage. What good looks like here: a clearly labelled key stored in a password manager, not pasted into a Slack message or shared document.

Step 2: Set a Usage Budget and Billing Limit

Before connecting anything, add a payment method to your OpenAI account and set a monthly spending cap under the billing settings. CV screening is a relatively light workload, typically a few thousand tokens per candidate, so most small and mid-sized hiring teams spend well under $20 a month even at meaningful volume.

Setting a hard cap protects against runaway costs from a misconfigured integration or an unexpectedly large CV batch. It also gives finance or operations stakeholders a concrete number to sign off on, which tends to smooth internal approval for hiring teams that need budget clearance before connecting a new billing account. What good looks like: a spending limit set slightly above your expected monthly usage, with an email alert configured at 80% of that cap.

Step 3: Locate the AI Integration Settings in Your ATS

Every ATS handles this slightly differently, but the pattern is consistent. In Klearskill, this lives under account settings, in a dedicated AI provider section where you can choose between using Klearskill's included AI or connecting your own key. Other platforms may call this "AI settings," "integrations," or "API connections."

Look for a field asking for an API key and, in some cases, a model selector. Some platforms separate this by workspace or team, so if your company runs multiple hiring workspaces, check whether the key needs to be added once globally or separately for each team. What good looks like: finding this setting without needing to contact support, since a well-designed ATS surfaces it clearly in account or workspace settings rather than burying it in a support article.

Step 4: Paste the API Key and Choose a Model

Paste your OpenAI API key into the field provided. Most platforms will validate the key immediately by making a small test call, so you should see a confirmation within a few seconds. If a model selector appears, choose a current general-purpose model suited to text classification and reasoning tasks, since CV screening involves comparing structured candidate data against job requirements rather than generating long-form content.

Avoid choosing the cheapest, smallest available model purely to save cost. CV screening accuracy depends heavily on the model's ability to reason about relevant experience, and a slightly higher per-token cost is usually worth it given how few tokens a single CV screening call actually uses. What good looks like: a validated connection showing a green "connected" status with no error messages.

Step 5: Run a Test Screening on a Real Job

Before rolling the integration out across live hiring, create a test job posting or use an existing one with a small batch of CVs, ideally five to ten, including a mix of strong and weak candidates you already have a judgment on. Run the screening and compare the AI's ranking against your own assessment.

This step catches configuration issues early: a mismatched job description, an overly narrow set of screening criteria, or a model that is being too lenient or too strict. It is worth deliberately including one or two CVs you know are weak matches alongside the strong ones, since a well-configured screening setup should rank them clearly lower rather than lumping every candidate into a similar score band. What good looks like: the AI's top-ranked candidates roughly matching who you would have shortlisted manually, with clear, specific reasoning attached to each score rather than vague justifications.

Step 6: Review the Screening Criteria and Adjust

Most ATS platforms let you define what the AI screens for beyond just matching keywords from the job description, including years of experience, specific skills, education requirements or industry background. Review these criteria against the test results from Step 5 and adjust anything that produced surprising rankings.

This is also the point to decide how much weight to give different factors. A senior technical role might weight specific tools and years of experience heavily, while a customer-facing role might weight communication indicators and relevant industry exposure more. It helps to write these criteria down in plain language, almost like a mini scorecard, before configuring anything in the ATS itself, since a clear written version makes it easier to spot gaps or bias in the criteria before they affect real candidates. What good looks like: screening criteria that reflect what actually matters for the role, documented so a second recruiter could review and understand the logic.

Step 7: Roll Out to Live Hiring and Monitor Usage

Once the test batch results look accurate, connect the integration to your active job postings and let CVs flow through automatically as candidates apply. Check your OpenAI usage dashboard weekly for the first month to confirm costs are tracking where you expected, and spot-check a sample of AI-screened candidates against your own judgment periodically.

Klearskill customers using their own OpenAI key at this stage are typically screening unlimited CVs on the Starter plan for $10 a month plus OpenAI's usage-based charges, which for most small hiring teams stays in the range of a few dollars monthly. Set a recurring reminder, weekly at first and then monthly once things stabilise, to glance at the OpenAI usage dashboard rather than relying on the billing cap alert as your only check-in point. What good looks like: consistent screening quality after 30 days, with no unexpected cost spikes and no drift in ranking accuracy.

Step 8: Revisit the Setup as Your Hiring Volume Grows

As hiring volume increases, periodically re-run the test process from Step 5 with a fresh batch of CVs to confirm accuracy holds at scale. Teams that outgrow the bring-your-own-key model, either because usage-based AI costs start exceeding a flat fee or because they want AI included without managing an external account, typically move to a plan like Klearskill's Pro tier, which includes AI at $50 a month with a 10,000 CV per month fair-use cap that lifts entirely if you connect your own key instead.

What good looks like: a documented decision point, whether that is a cost threshold or a headcount milestone, for when to reassess whether bring-your-own-key still makes sense for your team.

Common Pitfalls

Storing the API key insecurely. Pasting an API key into a shared document or unencrypted spreadsheet is a common and avoidable security lapse. Use a password manager or your ATS's dedicated secure field instead.

Skipping the test screening step. Connecting the key and going straight to live hiring without validating against known candidates means you will not catch a poorly configured screening criterion until it has already affected real applicants.

Setting no billing cap. An unexpected batch of CVs or a misconfigured automation can generate unusually high API usage. A billing cap and usage alert prevent this from becoming a surprise invoice.

Choosing the wrong model for the task. Using an overly minimal or outdated model to save a fraction of a cent per screen can measurably reduce screening accuracy, which costs far more in poor hiring decisions than it saves in API fees.

Forgetting to revisit criteria as roles change. Screening criteria set up for one role rarely transfers cleanly to a different role. Reviewing and adjusting per job type keeps accuracy consistent.

Treating the connection as set-and-forget. Job requirements, hiring managers' priorities and even the underlying OpenAI model can change over time. A quarterly check-in on screening accuracy, even a brief one, catches drift before it affects a full hiring cycle.

Tools That Help

OpenAI Platform is where the API key itself is generated and where usage and billing are monitored throughout the integration's life.

Klearskill is built specifically to make this connection simple, supporting OpenAI, Anthropic, Gemini and OpenRouter keys on its Starter plan, with unlimited jobs, unlimited candidates and unlimited CV storage regardless of which AI provider you connect.

A password manager, such as any standard business-grade option your company already uses, should store the API key rather than any shared document or chat tool.

AI Screening Benchmarks

Teams that connect their own OpenAI key to a dedicated CV screening tool can expect to screen candidates for a few cents each rather than a flat per-candidate vendor fee. Klearskill's AI screening engine delivers 97% accuracy against manually reviewed shortlists, with customers reporting a 92% reduction in time spent on manual CV review after switching. Recruiters using generative AI in their hiring workflow save roughly 20% of their working week, according to LinkedIn's Future of Recruiting report.

Frequently Asked Questions

Do I need coding skills to connect OpenAI to my ATS?

No. Most modern ATS platforms, including Klearskill, only require you to paste an API key into a settings field. No coding, scripting or technical integration work is needed for standard CV screening use cases.

How much does it cost to run CV screening through my own OpenAI key?

Costs vary by model and CV length, but most teams spend a few cents per candidate screened, meaning even a few hundred CVs a month typically costs a few dollars in OpenAI usage on top of your ATS subscription.

Is it safe to connect my OpenAI key to a third-party ATS?

Reputable ATS platforms store your API key encrypted and use it only to make screening calls on your behalf, never sharing it with other customers. Always check that your ATS explicitly states this in its security documentation before connecting a key.

What happens if I run out of OpenAI credit mid-screening?

Screening calls will simply fail until you add more credit or increase your spending cap, and most ATS platforms will flag the failed screenings clearly rather than silently skipping candidates. Set a billing alert well before you expect to hit your cap to avoid this.

Can I switch between my own OpenAI key and an included AI plan later?

Yes. On Klearskill, you can switch from Starter's bring-your-own-key model to Pro's included AI at any time, and vice versa, without losing any job postings, candidate data or screening history.

Which OpenAI model is best for CV screening specifically?

A current general-purpose model with strong reasoning capability is typically best for CV screening, since the task involves comparing structured experience against job requirements rather than generating creative text. Avoid the smallest, cheapest models, as screening accuracy can suffer.

Does connecting my own OpenAI key remove Pro's CV screening cap?

Yes, on Klearskill specifically. Pro's included AI comes with a 10,000 CV per month fair-use cap, but connecting your own OpenAI, Anthropic, Gemini or OpenRouter key removes that cap entirely while keeping the rest of Pro's features, including unlimited users on your company domain.

Stop Screening CVs Manually in 2026

Connecting your own AI key is the fastest way to get accurate, affordable CV screening without waiting on a vendor's bundled AI pricing. Klearskill delivers 97% AI screening accuracy and a 92% reduction in time spent screening, with two plans, both unlimited on jobs and candidates: Starter at $10 a month with your own AI key, or Pro at $50 a month with AI included. Try Klearskill free with 1 job and 25 CVs screened, no card needed.