A traditional applicant tracking system (ATS) helps you manage hiring steps—post a job, track applicants, move them through stages. A well-rounded ATS for mid-market or enterprise businesses (like Pinpoint) will help you manage your entire hiring process end-to-end, often with AI features. For example, Pinpoint offers features like AI Fields and AI Chatbot within the ATS.

The best AI applicant tracking systems help you make faster, more consistent decisions. AI and automation features can help you do things like:

• Assess applications and resumes for role fit
• Turn broad inputs into structured candidate data
• Automate workflows
• Generate text or responses

But they can’t make smart decisions unless your prompts are smart, too.

When people talk about AI in recruitment, they’re usually thinking of what Large Language Models (LLMs) can do, within the tech they use for hiring. Things like:

• Reading a resume
• Summarizing candidate experience
• Screening for specific skills
• Writing emails or interview questions

That’s all powered by LLMs; a type of AI trained to understand and generate natural language. And LLMs don’t think like people. They don’t know what a strong candidate looks like. They interpret your prompts literally. So if your prompt is something like:

“Experience in retail”

The LLM won’t have enough context. Does that mean any kind of retail? Does six weeks over Christmas count? Is this front-of-house or back-office?

This is where the problems start. If your instruction is unclear, the output will be too.

Whether you’re evaluating experience, skills, or suitability, every AI prompt should be:

Clear

Avoid generic labels. Be specific and concrete.

“Retail experience” is not specific.

“The candidate has worked in a customer-facing retail role for at least 6 months” is.

Contextual

Spell out what does and doesn’t count.

“The candidate has worked in a commercial kitchen as part of a paid role. School placements or volunteer shifts don’t count.”

In Pinpoint, recruiters can create AI-powered fields for candidate evaluation. This means you can score candidates against a job description, flag specific experience, or summarize skills automatically, turning long applications into structured data you can sort and search. It’s especially important when prompting for this type of AI Field that it’s answerable with “yes” or “no”. Prompts that lead to binary answers are usually easy to automate, filter, and action in your ATS.

Let’s walk through how to improve a prompt for Pinpoint’s AI Fields feature step-by-step.

❌ Bad prompt

“Hospitality”

Too broad. Does that include bartending, reception, event staffing? Does one shift count?

⚠️ Better prompt

“The candidate has hospitality experience.”

Still vague. Doesn’t define the kind of work, or what counts as experience.

✅ Great prompt

“The candidate has worked in a hospitality role in a paid job. For example, front desk at a hotel or serving in a restaurant. Volunteering or school placements don’t count.”

This prompt:

• Defines paid experience
• Gives examples
• Sets a clear standard

And most importantly, it’s something the AI Fields feature can confidently evaluate with “yes” or “no.”

Here are some more examples of clear, contextual AI prompts:

Weak prompt: “Retail”
Improved version: The candidate has worked in a customer-facing retail role for at least six months. Excludes roles with no direct customer interaction or any experience shorter than one month.

Weak prompt: “Hospitality experience”
Improved version: The candidate has held a paid role in a hospitality setting—such as a hotel front desk or waiting tables. Excludes short-term volunteering or one-off event support unless explicitly stated.

Weak prompt: “Warehouse”
Improved version: The candidate has worked in a warehouse environment and used inventory or scanning systems. Excludes general labor roles without stock handling or system use.

Weak prompt: “Good with customers”
Improved version: The candidate has handled customer requests, feedback, or complaints as part of a paid role. Excludes indirect or back-office exposure unless they directly contributed to service outcomes.

Weak prompt: “Leadership”
Improved version: The candidate has directly led or supervised a team of three or more people in a paid role (e.g., shift lead or team supervisor). Excludes informal mentoring unless they were formally accountable for performance or outcomes.

AI in recruitment only delivers value when inputs are accurate. Before you hit “save” on a prompt, ask yourself:

“If I had to answer this question about someone’s CV, would I know exactly what to look for?”

If not, it’s too vague. Rewrite it until the meaning is obvious.