AI Recruiting Tools for Small Businesses: What's Worth Paying For

For most small businesses, only two things are worth paying for: a general AI assistant at about $20 a month for adverts, scorecards and interview questions, and, once roles regularly draw 50 or more applicants or you hire more than five people a year, an applicant tracking system with AI screening. Skip AI video-interview scoring and personality analysis.

The trap is paying for recruiting software sized for an agency when your real problem is one of three things: adverts that attract the wrong people, a pile of applications nobody has time to read, or candidates who vanish because nobody replied for a week. Each needs a different tool, and the cheapest fix for the first is often a better-written advert rather than any software at all. Below are nine kinds of tool, what they cost as of September 2026, and a verdict on each.

Follow me on Instagram@sagnikteaches

1. A general AI assistant for adverts, scorecards and questions

Verdict: worth paying for. ChatGPT Plus and Claude Pro are $20 a month each (Claude Pro is $17 a month billed annually); a business plan such as ChatGPT Business or Claude Team costs $25 a seat monthly and needs at least two seats. A business plan is the better home for candidate information, because business content isn't used for model training by default.

Connect on LinkedInSagnik Bhattacharya

This is where AI saves a small employer the most time per dollar: turning a vague idea of the role into a clear advert, a scorecard that says what "good" looks like, and interview questions tied to that scorecard. An illustrative prompt for a landscaping firm hiring for the spring:

Subscribe on YouTube@codingliquids

Write a job advert for a seasonal landscaping operative, March to
October, 40 hours a week, some Saturdays in summer. Duties: mowing,
hedge cutting, planting, loading the van, following the team leader's
job sheet. Must be able to lift 25 kg and work outdoors in all
weather. A driving licence is useful but not required. We train on
all machinery. Pay: [rate] an hour.
Keep it under 250 words, plain English, no clichés. Then write a
5-point scorecard for shortlisting and 6 interview questions that
test the scorecard points.

What came back, shortened (illustrative):

Scorecard (score each 0-2)
1. Reliability: evidence of turning up on time for shift or seasonal work
2. Physical outdoor work: has done manual work outside before
3. Following instructions: can work from a written job sheet
4. Safety awareness: mentions or asks about safe machinery use
5. Availability: can do March-October including some Saturdays

Advert opening
"Enjoy working outside? We're a friendly, young team looking for an
energetic landscaping operative to join us this spring..."

The scorecard is useful as it stands. The advert needs two fixes. "Young team" and "energetic" can read as a preference for younger applicants, which is exactly the kind of wording that invites discrimination complaints; replace them with what the job involves ("a team of five, working outdoors on residential gardens"). And the draft dropped the line "we train on all machinery", which is the sentence most likely to attract people without landscaping experience. Always check an AI-written advert against your own notes line by line.

2. An applicant tracking system with AI screening

Verdict: worth paying for once volume justifies it. An applicant tracking system (ATS) collects applications from job boards into one list, tracks each candidate's stage and sends emails. The AI part usually summarises CVs against your criteria, suggests a match score and drafts replies. Sample list prices:

Where the threshold sits is simple arithmetic. If reading and replying to each application takes three minutes, 60 applications for one role is three hours. At five hires a year with 60 applicants each, that's 15 hours of reading plus the chasing. An ATS at $15-$20 a user a month earns its place at that volume just by keeping everything in one list; at $299 a month, you want double-digit hires a year or several roles open at once.

Take an illustrative cleaning company that hires about 25 domestic cleaners a year because of natural turnover, with 60-80 applicants per advert. Before an ATS, applications arrived by email, text and three job boards, and the owner lost track of who had been called. With Manatal's Professional plan for two users (about $30 a month annually), every application lands in one list, the AI summary flags who meets the must-haves (weekday mornings, own transport, cleaning experience), and templated emails go out the same day. The owner still reads every shortlisted application herself. The yearly cost is about $360, against the 60 or so hours of sorting and chasing it replaces.

Credits deserve a quick sum before you choose a credit-based plan. The cleaning company's 25 roles a year at 70 applicants each means 1,750 AI evaluations. On Workable that would be 1,750 credits from the 3,000 included, leaving room for some sourcing (2 credits a person) but only about 125 AI candidate chats at 10 credits each. If the plan is to let the AI agent chat with every applicant, the allowance runs out in months, and extra credits cost $0.10-$0.12 each in packs that expire after a year. Work out which AI actions you'd actually use, multiply by your volume, and compare that with the allowance before signing.

For a deeper look at judging these features in a demo, see how to evaluate the AI features in recruitment software, and if you already have an ATS, whether to bolt AI screening on or switch.

3. Job-board AI for finding people who aren't applying

Verdict: worth it only for hard-to-fill roles. Indeed launched Talent Scout, an AI agent that searches candidate profiles, surfaces matches and helps draft outreach, alongside its Smart Sourcing subscription for searching CVs and contacting people directly. LinkedIn's Recruiter Lite is aimed at people who hire "just a few people a year", with 30 InMail messages a month and search across your first, second and third-degree network. Prices vary by market and change often, so check them in your own employer account.

These tools matter when the right people aren't looking at adverts. An illustrative electrician's firm needing a qualified, experienced electrician with a particular certification might advertise for six weeks and get no suitable applicants. Paying for a month of sourcing to message 30 qualified people directly can fill the role; paying for it all year to hire a trainee who would have applied anyway is waste. Buy sourcing by the month, for a named role, and cancel when it's filled.

The outreach message decides whether sourcing works, and AI-drafted messages tend to sound like every other recruiter's. A before-and-after for that electrician's role. The AI draft: "Hi [name], I came across your impressive profile and think you'd be a great fit for an exciting opportunity with our growing team." The edited version: "Hi [name], I run a seven-person electrical firm and need an approved electrician for commercial fit-outs, mostly within 15 miles, no overnight stays. Your inspection and testing experience is exactly what we're short of. Pay is [range]; van and tools provided. Would a ten-minute call this week be worth it to you?" The second names the work, the pay and the reason you wrote to them. Give the assistant those facts and ask for a message under 80 words with no flattery, and the first draft gets much closer.

4. AI-written screening questions and knockout forms

Verdict: cheap and worth doing. Most applicant tracking systems and job boards let you add screening questions to the application. An assistant can draft them from your scorecard, and a few well-chosen questions cut a pile of 80 applications to the 20 worth reading. A filled-in example for the cleaning company above:

Notice the "how it's used" column. Knockout questions that automatically reject people are where small employers get into trouble. Picture the same company's first attempt: a question "Do you have your own car?" set to auto-reject anyone answering no. It quietly rejected a dozen applicants living on bus routes the company served well, including two with years of cleaning experience. Changing it to a flag rather than a rejection fixed it. Keep automatic rejection for genuine legal or safety requirements only, and review a sample of rejected applications every month.

5. Interview scheduling that books itself

Verdict: worth it at volume; usually already included. The slowest part of small-business hiring is often the back-and-forth to find an interview time, and good candidates accept other offers while you're emailing. Most applicant tracking systems include self-booking: the candidate picks a slot from your calendar. If you use a separate booking tool, the free tier of a common scheduling app is often enough for occasional interviews.

A quick before-and-after from a roofing contractor hiring two labourers. Before: the owner phoned shortlisted applicants from site, left voicemails, and took four days on average to fix interviews; three of eight shortlisted people had taken other jobs by then. After: shortlisted applicants got an email the same evening with a booking link for two fixed interview windows a week. Six of eight booked within 24 hours. No AI was needed for that part; the saving came from removing the phone tag.

6. AI note-takers for interviews

Verdict: useful, with consent and care. Meeting note-takers such as Otter.ai (Pro $8.33 a user a month billed annually, $16.99 monthly) transcribe a video interview and draft a summary. That lets the interviewer listen instead of scribbling, and it makes comparing candidates fairer, because every interview gets the same record.

Tell the candidate at the start, ask if they're comfortable, and offer to switch it off. Then summarise against the scorecard, not in general. An illustrative summary format worth asking for:

Candidate: [initials]   Role: Landscaping operative
Scorecard point      Evidence from interview             Score
Reliability          3 summers at a garden centre, no     2
                     mention of absences; left for college
Outdoor work         Yes, as above; also farm work        2
Following job sheet  Described using a stock list; no     1
                     written instructions example
Safety               Asked about PPE and training         2
Availability         Can't do Saturdays in August         1
Questions the candidate asked: training on the mower, pay rise
at end of season.

Check the summary against your memory the same day. Transcription tools can mishear names, qualifications and numbers, and an AI summary may state a conclusion ("strong candidate") the interview doesn't support. Delete recordings when the hiring decision is made unless you have a clear reason and notice to keep them.

Settings to check the day you switch AI screening on

Most of the risk in an AI-enabled ATS sits in its default settings, not in the model. Before the first advert goes live, go through these:

• Automatic rejection: off, or limited to genuine legal requirements such as a licence the job cannot be done without.
• Match scores: find out what the score is based on (your criteria, the job description, or the vendor's own model) and whether you can see why a candidate scored as they did.
• Who sees AI summaries: the hiring manager, not everyone with a login.
• Candidate data retention: set a period after which unsuccessful applications are deleted, and say so in your privacy notice.
• Candidate-facing AI: if an AI chat or email agent talks to applicants, read five of its messages before letting it run, and make sure it says it is automated.

Wording for the privacy notice or advert footer, adapted from what I'd suggest to a small employer (have your adviser check it):

How we use your application: We use recruitment software that
includes AI features to organise applications and summarise them
against the requirements in this advert. A member of our team reads
every shortlisted application and makes every decision. If you would
like a person to review your application without the AI summary,
reply to our acknowledgement email and say so. We keep applications
for [6] months after the role is filled, then delete them.

7. Skills tests and assessments

Verdict: sometimes, for testable office skills. Some applicant tracking systems sell assessments as an add-on (Workable's Assessments+ is $59 a month on its Standard plan), with AI marking of written answers. For an office role where the skill can be tested on screen, such as a scheduler who must use spreadsheets or write clear customer emails, a 20-minute test before interview saves both sides time.

For trades and hands-on work, an online test measures the wrong thing. A roofing contractor learns more from a paid half-day trial on a simple job, with the team leader's notes against the scorecard, than from any online assessment. A useful middle ground for an office role: have an assistant draft a short, realistic task (for example, "reply to this customer who is unhappy that their appointment moved"), mark it yourself against three points, and skip the add-on entirely. An illustrative task and marking guide for a cleaning company's office coordinator:

Task: A regular client emails at 7.45am: "Nobody turned up
yesterday and I took the morning off work for this." Write your reply.
Mark (0-2 each):
- Apologises without excuses and takes ownership
- Offers a concrete fix (new slot today or tomorrow, named cleaner)
- Checks what went wrong internally before promising anything else

8. Drafting offers, rejections and reference requests

Verdict: worth it, and covered by the assistant from item 1. A prompt such as "write a kind, brief rejection for a candidate we interviewed who was good but lacked experience with irrigation systems; encourage them to reapply next season" produces a better note than most rushed owners write, and candidates remember how they were treated. The same goes for offer letters (have your employment adviser approve the template once, then let AI fill in the details) and reference requests with specific questions tied to the scorecard.

A before-and-after rejection makes the point. Before: "Thanks for applying, unfortunately you were unsuccessful." After, edited from an AI draft: "Thank you for coming in on Tuesday. We've offered the role to someone with more irrigation experience, which this job needs from day one. You came across as reliable and keen, and we'd be glad to hear from you when we advertise again in February." It takes under a minute with AI, and it keeps a possible future hire warm.

9. Onboarding packs and first-week questions

Verdict: worth it once you hire regularly. Hiring doesn't end at the offer. An assistant can turn your scattered notes into a first-week checklist, a plain-English summary of the staff handbook and a list of who to ask about what. A company that hires 25 cleaners a year repeats the same first-week questions 25 times; putting the answers in a shared document or a staff help assistant saves the owner's phone from ringing every evening. The tutorial on answering staff HR questions with an AI helpdesk covers the next step.

Three kinds of AI hiring tool I'd skip

• AI video-interview scoring. Tools that score recorded answers on word choice, tone or facial expression are hard to validate, easy to challenge, and aimed at employers screening thousands. A small business gains nothing a phone call wouldn't give it.
• Personality or "culture fit" prediction from CVs or social profiles. These claims are rarely backed by evidence you can check, and "fit" is where bias hides.
• Fully automatic rejection or ranking. Letting software decide who is out, with no person reviewing, carries legal and reputational risk for very little time saved at small volumes.

There is a regulatory reason too. If you recruit in the EU, AI used for recruitment and selection is listed as high-risk under the EU AI Act; those obligations have been deferred to 2 December 2027 for stand-alone systems, but vendors and employers are already preparing for them. Where data-protection law such as the GDPR applies, candidates have rights around decisions made solely by automated means. Keeping a person in every decision, and a note of why, keeps you on the right side of both. For employment law questions specific to your business, ask an employment adviser.

Checking the AI isn't filtering out good people

Before trusting any AI match score, test it on people you already know. A 30-minute check:

1. Gather 15-20 past applications for a role, including the people you hired and a few who turned out well elsewhere.
2. Run them through the tool's screening with your current criteria.
3. Look at where your good hires landed. If your best cleaner of the last three years sits in the bottom third, the criteria or the tool are wrong.
4. Check for patterns: are applicants with career gaps, non-standard job titles or CVs written in a second language scored lower for no job-related reason?
5. Write down what you found and what you changed. That note is also your evidence of a fair process if a decision is ever questioned.

A realistic example of what this catches: a small removals firm found its tool scored applicants with "porter" or "warehouse operative" on their CVs far below those with "removals", although the work is almost identical. Adding those job titles to the criteria moved three strong past hires back into the top group. The fuller method is in bias checks for AI CV screening.

What a sensible hiring stack costs at three sizes

Illustrative yearly costs using the list prices above; your mix will differ.

The plumbing firm's line is worth a second look. With one or two hires a year, the money is better spent on a well-written advert in the right place than on software. The landscaper's pain is volume in a six-week window every spring, so a per-user ATS switched on for those months (check whether the vendor allows pausing) and a booking link solve it. The cleaning company has continuous recruitment, so a proper system with texting pays, because candidates for hourly roles answer texts far faster than emails.

Where to spend the first $100

If you do one thing this month: pay for one assistant, and use it to write a scorecard and advert for the next role you'll fill. If applications then arrive faster than you can read them, add an applicant tracking system at the cheapest per-user tier and set screening questions that flag rather than reject. Only after that consider sourcing credits or add-ons, and only for a named role. The step-by-step process that ties these together is in setting up an AI-assisted hiring process for a small team.

Questions about paying for AI hiring tools

Can I use ChatGPT or Claude to screen CVs without an ATS?

For a handful of applicants, you can paste anonymised CVs into a business-plan assistant with a scorecard and ask for a summary against each criterion. Don't ask it to rank or reject people, keep your own notes on every decision, and check your privacy notice tells candidates how their data is handled. Above about 30 applicants a role, an applicant tracking system is less work and keeps a proper record.

Do I have to tell candidates that AI is used in hiring?

It is good practice everywhere and increasingly a legal expectation. Say in the job advert or privacy notice which steps use AI, that a person makes every decision, and how to ask for a human review. Where data-protection law such as the GDPR applies, candidates have rights around decisions made solely by automated means. Ask an employment adviser what applies to you.

Are free applicant tracking systems good enough?

For one role at a time, often yes. Breezy's free Bootstrap plan, for example, allows one active position with unlimited users and candidates from the last 30 days. Free plans usually leave out AI screening and cap how many jobs you can run at once, so they suit a business hiring two or three people a year rather than running rolling recruitment.

What is an AI credit in recruiting software?

A unit the vendor charges for each AI action. In Workable, for example, evaluating one candidate uses one credit, finding a passive candidate uses two and an AI chat with a candidate uses ten. Plans include an allowance and you buy more in packs. Estimate your usage from applicant numbers before choosing a plan, because credit packs expire.

Further reads

• How to Keep Customer Data Private When Your Team Uses AI — Keeping candidate and customer data private when staff use AI.
• AI or a New Hire? How to Decide Before You Recruit — Checking the role needs a person before you recruit at all.
• AI Bias in Small Business Decisions: Hiring, Pricing and Credit — The wider picture on bias in AI-assisted decisions.
• Questions to Ask a Recruitment Agency About Its AI Screening — If an agency does the screening for you instead.
• Shadow AI: How to Stop Staff Pasting Client Data Into Free Tools — Stopping staff pasting candidate CVs into free AI tools.
• How to Write Job Descriptions With AI That Attract Good Hires — AI writes fluent job adverts; good candidates respond to specifics. The inputs, prompt and eight edits that turn a generic draft into one worth applying to.
• Can AI Screen CVs Fairly? What Small Employers Need to Know — How AI CV screening goes unfair, what the rules expect of small employers, and a six-step set-up with tests you can run in an afternoon.
• How to Write a Job Advert for an AI-Savvy Admin Assistant — A bakery's admin advert built line by line: which AI skills to ask for, wording that attracts the wrong people, and a task that shows who checks their work.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: Workable, Manatal and Breezy HR pricing pages; Indeed announcement of Talent Scout and its Smart Sourcing page; LinkedIn Recruiter Lite page; facts sheet for chat assistant prices, Otter.ai pricing and EU AI Act dates. Checked September 2026.