Skip to content

AI features for every stage of recruitment

Recruiting work splits into finding people, judging fit, and keeping the pipeline moving.

This page explains which of those tasks the AI toolkit covers, what each tool takes as input and returns, and where a human decision still belongs. It is written for agency recruiters and talent teams already working inside an ATS who want to know whether embedded AI removes real steps — or just adds another tab.

What the AI toolkit actually does

The capabilities fall into four practical groups. Each one is tied to a record type you already keep in the ATS, which is why the output can be acted on without exporting data.

  • Sourcing and matching. You describe a role in plain language, in Boolean, or by pasting job requirements. The tool returns candidates from your talent pool ranked by skills, experience and role fit.
  • Profile understanding. Resumes and candidate records are turned into structured, summarizable information so a recruiter can read the essentials quickly.
  • Written output. Email replies and outreach content can be drafted with the candidate, client and job context already attached.
  • External assistant access. Through MCP, assistants such as Claude, ChatGPT and Perplexity can query your recruitment data in natural-language conversations.

A useful way to read this list: the first three groups keep work inside the ATS, while the fourth deliberately opens the ATS to a tool you already use.

Search inputs and what comes back

Search is the clearest example of how the AI layer changes day-to-day work. There are three input styles, and they suit different moments.

  1. Natural language — best when you are exploring. You write the requirement the way you would say it to a colleague.
  2. Boolean logic — best when you need precision and repeatability, for example a saved string for a niche skill combination.
  3. Job requirements — best when a role is already defined; the requirement set becomes the query.

In all three cases the output is a set of candidates judged on skills, experience and role fit. The tool narrows the pool; it does not make the placement decision. Treat the returned shortlist as a first pass that a recruiter reviews against client context, availability and compensation expectations — none of which the search input itself captures.

A worked example: one role, three AI steps

Suppose a client brief asks for a mid-level engineer with a specific stack and several years in a regulated environment.

Step 1 — Find. Enter the requirement as a natural-language sentence, or switch to Boolean if the stack terms need exact operators. The search runs across the existing talent pool and returns candidates whose skills, experience and role fit align with the brief.

Step 2 — Read. For the strongest matches, generate a profile summary. This compresses the resume and record into the points a recruiter needs before a screening call, instead of re-reading the full document.

Step 3 — Reply. Draft the outreach or reply with the candidate, client and job context already in place. The recruiter edits tone and specifics, then sends.

The value is not that any single step is impossible without AI. It is that the three steps share one context, so information is not retyped between a search tool, a document and an inbox.

Embedded AI vs your own assistant vs a standalone chatbot

These three setups are often confused, and the difference matters when you choose a workflow.

  • Embedded AI lives inside the platform. It works on candidates, clients, jobs and workflows already in the ATS, and no separate subscription or export is needed to use it.
  • Your own assistant via MCP connects tools like Claude, ChatGPT or Perplexity to your recruitment data where MCP is available on your plan. You keep the assistant you prefer and ask questions in natural language; the assistant reads from your ATS context.
  • A standalone chatbot has no access to your records. You paste information in and copy results out, and the model does not know your pipeline, your clients or your job history.

If your team already has a preferred assistant, the MCP route is the one that preserves it. If you want the fewest moving parts, the embedded features cover search, matching, summaries and drafting without leaving the platform.

Data handling and plan availability

Two checks belong before any rollout.

Data use. The platform's stated policy is that recruitment records stay private and are used to deliver AI functionality within the account, and that candidate, client and job data are not used to train public AI models. Check that statement against your own client contracts and any data-processing terms you have signed before relying on it.

Plan coverage. Not every AI capability is on every plan. Availability depends on your subscription tier, so check the current plan breakdown or ask the team which capabilities fit your workflow before you build processes around them.

A short pre-rollout checklist:

  • [ ] Confirm which AI features your plan includes.
  • [ ] Confirm your data-processing terms cover AI-assisted processing.
  • [ ] Pick one role and run the three-step example end to end.
  • [ ] Decide whether recruiters use embedded AI, an MCP-connected assistant, or both.
  • [ ] Agree who reviews AI-drafted content before it reaches a candidate or client.

Frequently asked questions

Is the AI built into the ATS, or do I need separate AI tools?

The AI capabilities are part of the platform itself, so recruiters can use them while managing candidates, clients, jobs and workflows. MCP is an optional bridge that connects the ATS to external assistants; the core features are built in and do not depend on that bridge. Check your subscription or plan terms before relying on them.

Will my candidate and client data be used to train AI models?

The platform's stated policy is that recruitment data stays private and is used only to provide AI functionality inside your account, and that candidate, client and job data are not used to train public AI models. Check that statement against your own data-processing terms.

Does the AI work with the candidate database I already have?

Yes. The features operate on data already in the ATS. You can search the database, match candidates, summarize profiles and create content, which means existing records become more useful rather than needing to be rebuilt.

How is this different from using ChatGPT on its own?

A general chatbot has no view of your pipeline. Here the AI draws on the candidate, client, job and workflow records already held in the ATS, so a recruiter does not have to shuttle information between a chat window and the database to get a usable answer.

Can I connect my own AI assistant?

Yes, where MCP is available on your plan. MCP lets you connect assistants such as Claude, ChatGPT and Perplexity to your recruitment data. You can ask questions, look things up and work with ATS data through natural-language conversation.

Which AI features are included in my plan?

It depends on your subscription. Check the current pricing breakdown for the feature list per plan, or speak with the team about which AI capabilities suit your workflow. Treat any feature list as plan-specific rather than universal.