Job descriptions: how to ask for AI skills your small business actually needs
By the Redstone HR team · Published
Subscribe to our newsletter
Read about our privacy policy.
To put AI skills into job descriptions well, name the everyday tasks the person will use AI tools for, say how they check the results, and mark the skill as essential only if the role cannot be done without it. "Must have AI skills" on its own tells applicants nothing. A line such as "uses AI tools to draft letters and summarise documents, then checks them for accuracy" tells the right people they fit. This guide is for office managers, ops leads and founders who write job descriptions alongside everything else.
Why AI skills are turning up in job descriptions, and why vague wording backfires
AI skills are appearing in job descriptions because small businesses want people who can use AI tools sensibly. Vague wording backfires because it attracts the wrong applicants and puts off the right ones. A confident applicant who has never called themselves "AI-skilled" may not apply. Someone who lists every tool they have tried may apply for a role that really needs care with personal data and a good eye for errors.
The cause is usually how job descriptions get written in a small business. There is no HR department, so the office manager or founder copies a template, adapts last year's version and adds new requirements as a line at the bottom. "AI skills" gets added as a buzzword in the same way "proficient in Microsoft Office" once did.
Aim for a job description that says three things plainly:
- what the person will do with AI tools
- at what level
- why it matters to the role
What a job description is, and what it is not (UK)
A job description sets out a role's purpose, main tasks, responsibilities and the skills or qualifications needed, in plain language. It often comes with a person specification, which lists the skills, knowledge and experience you are looking for.
A job description is not the same as two documents it is often confused with:
- The job advert is the marketing version. It is shorter and written to get people interested.
- The written terms of employment (the written statement of employment particulars in the UK) cover pay, hours, holiday entitlement, bank holidays and other terms. That is a contractual document, not a description of the work.
The difference matters for AI skills. AI skills belong in the tasks and the person specification, where you can say what the person does and how well. A slogan such as "AI-first team" in the advert's headline adds nothing a candidate can judge themselves against.
Does this role really need AI skills? A quick test
Most small-business roles only need AI skills named if specific weekly tasks would be done faster or better with a tool. Before adding anything, ask three questions:
- Which tasks repeat every week? Drafting letters, answering the same staff questions, summarising documents, preparing monthly figures.
- Where would a tool save time? First drafts and summaries are common answers. Decisions about people are not.
- Who checks the result? If no one does, the risk sits with whoever you hire, and the job description should say they are expected to check.
Roles where AI tools often fit include office manager, operations coordinator, practice manager, and marketing or admin support. These roles involve a lot of writing, reading and answering questions.
Roles where AI skills usually don't need to appear include hands-on, shift-based and site roles: a dental nurse, a care worker, a site operative. Leave AI skills out of those job descriptions rather than padding them.
Approach What happens Better for Add "AI skills" to every role Job descriptions read alike; candidates can't tell what is actually needed; some good applicants self-select out Rarely the right choice Name AI skills only where a task needs them Each job description describes real work; candidates can judge their fit; interviews have something concrete to test Most small businesses
Which AI skills to name: tasks, not buzzwords
The AI skills worth naming in job descriptions are specific everyday tasks plus the judgement to check the output. In small organisations, admin drifts across spreadsheets, email threads and paper forms, and one person ends up checking, chasing and answering the same questions. The useful skills are the ones that reduce that load safely.
Task-based wording you can adapt:
- Drafting and summarising: "Uses AI tools to draft first versions of letters, emails and notices, and to summarise long documents, then checks them for accuracy and tone before sending."
- Answering staff questions from written policies: "Answers staff questions about annual leave, carrying holiday over and time off using our written policies, and knows when a question needs escalating to a manager."
- Checking output: "Spots when an AI answer is wrong, incomplete or out of date, and checks it against the source document."
- Protecting data: "Keeps personal data private and does not paste sensitive staff, client or patient information into public AI tools."
Avoid listing named tools unless the role truly depends on one. Skills move between tools, and brand names in a job description quickly go out of date. "Confident using AI writing and summarising tools" will still make sense in two years.
Essential or desirable? How to word AI requirements fairly
AI skills should usually be marked desirable rather than essential, because many capable candidates will learn them on the job. Make an AI skill essential only if the person could not do the role without it from day one.
Keep the wording plain and inclusive. Phrases such as "digital native" or "young, tech-savvy team" may imply an age preference, and age is a protected characteristic under the Equality Act 2010. Describe the skill, not the type of person.
Describe the level instead of asking for years of experience, because AI tools are too new for "three years' experience" to mean much. A simple scale works:
- Comfortable using: can use an AI tool for drafts and summaries with guidance.
- Confident using and checking: uses tools independently and reliably catches errors.
- Confident setting up and checking: can set up how the team uses a tool, write guidance and check others' use.
If you also hire outside the UK, local rules on job descriptions and hiring differ. Redstone HR's guides to Iowa labour laws for employers and affirmative action plan compliance show how much US requirements vary from state to state and employer to employer.
Example wording for an office manager or ops lead role
A strong office manager job description pairs real admin tasks with specific, checked use of AI tools. Here is the difference in practice.
Weak version:
Must have AI skills. Experience with ChatGPT essential.
Strong version:
Responsibilities Handle holiday requests and keep one accurate record of annual leave and holiday entitlement for all staff. Prepare monthly absence figures for payroll and check them before they are sent. Use AI tools to draft letters, policies and staff notices, and to summarise documents, checking every draft for accuracy before it goes out. Answer staff questions about time off and bank holidays from our written policies, and escalate genuine exceptions to the managing director. Person specification Desirable: confident using and checking AI writing and summarising tools. Essential: careful handling of personal data, including knowing what must not go into public AI tools.
The strong version tells candidates what the work is, where AI fits and where human judgement stays in charge. It is also designed to cut repetitive work: when staff can find answers to routine questions themselves, the person in the role spends their time on genuine exceptions rather than answering "how many days have I got left?" for the tenth time that week.
Don't forget cover: job descriptions, time off and capacity
A job description should say who covers the role during annual leave and bank holidays, and in a small business that line is often missing. When the office manager is off, someone still needs to approve holiday requests, answer staff questions and get figures to payroll.
Capacity planning in a small business means checking you will have enough people in each role on a given day once annual leave, bank holidays and sickness are counted. It is mostly about the next few weeks, not long-range workforce strategy. Adding a cover line to each job description, such as "Covers for the operations coordinator during their time off, and is covered by them in turn", makes those gaps visible before they happen.
For site-based teams the same problem plays out across crews and job sites; see how leave and cover work across construction crews.
Cover is also where tools help most. The usual failure is approving a holiday request without seeing who else is off, then finding out on a client day that you are short-staffed. A shared team calendar with coverage warnings at approval time, as in Redstone HR, lets whoever approves leave see who else is out before saying yes. A spreadsheet can record leave, but it cannot warn you about a gap at the moment you approve.
After the hire: setting the new starter up in week one
Good onboarding for a small business without HR means preparing before day one, booking check-ins at fixed points, giving the new starter one place to find practical answers, and setting up admin such as holiday entitlement in the first week.
If you hired partly for AI skills, give the new starter clear guidance in week one:
- Which AI tools the business uses, and for which tasks.
- Which data must stay private and never go into a public tool.
- Who checks AI-drafted work, and when.
Self-service makes week one easier for everyone. When employees can look up their own leave balances and ask policy questions themselves, the hiring manager is not pulled away for the same questions. In Redstone HR, employees check their balances and ask an AI policy assistant questions such as whether they can carry holiday over, so the person in charge handles only the exceptions.
Self-assessment: is your job description ready to publish?
A job description that asks for AI skills is ready to publish when you can answer yes to each of these:
- Does each AI requirement name a task, not just a tool or a buzzword?
- Is each AI requirement correctly marked as essential or desirable?
- Does the job description say how the person checks AI output and protects personal data?
- Is it written in plain English, free of age-coded or exclusionary phrases such as "digital native"?
- Does it say who covers the role during time off, and is holiday entitlement set up in week one?
Work through these five questions against the job descriptions you already have, not just the next one you write. If the role you are hiring for involves covering absences, use the free Absence Cost Calculator to see what gaps in cover cost your small business.
Frequently asked questions
Should I put AI skills in every job description?
No. Name AI skills only in job descriptions where specific weekly tasks, such as drafting letters or summarising documents, would be done better with a tool, and leave them out of hands-on, shift-based or site roles.
Should AI skills be essential or desirable in a job description?
Mark AI skills as desirable unless the role cannot be done without them from day one, because many capable candidates will learn to use AI tools on the job.
Should I name specific AI tools in a job description?
Only if the role truly depends on one tool. Describing the task, such as using AI tools to draft and summarise then checking the result, stays accurate when tools change.
What is the difference between a job description and the written terms of employment?
A job description sets out a role's purpose, tasks and required skills, while the written terms of employment cover pay, hours, holiday entitlement and bank holidays, so AI skills belong in the job description and person specification.
Should a job description say who covers the role during annual leave?
Yes. Naming who covers the role during annual leave and bank holidays makes cover gaps visible before they leave a small business short-staffed.
Keep reading
HR agent or AI assistant? What a small business really needs to manage annual leave
HR agent or AI assistant for annual leave? What each does, where they fall short for a small business, and how to stop approving holiday blind.
What are software vendors? The questions a small business should ask HR software vendors before buying leave software
What are software vendors, and what should a small business ask before buying leave software? The questions that expose cost, cover and set-up gaps.
HR Capacity Planning for Small Businesses: How to Stop Overlapping Leave Leaving You Short-Staffed
HR capacity planning for small businesses: set minimum cover, see who is off before approving annual leave and stop being caught short on busy days.
Onboarding Strategies for Small Businesses: A First 90 Days Plan That Doesn't Forget Time Off
Onboarding strategies for small businesses without HR: a practical first 90 days plan, including how to set up a new starter's holiday entitlement.
Free tools: Holiday entitlement calculator · Bradford Factor calculator · Free leave templates · Leave law guides · HR glossary
