The Talent Matching assistant is currently in beta and free to use. Availability and pricing may change, and information in this article may not reflect the most up-to-date version of the product.
The Talent Matching assistant is a conversational tool on the Talent Matching page that turns a plain-language description of the role into a working setup. Describe the kind of talent you're looking for, and the assistant drafts a calibration and a set of filters for you to review, so you don't have to piece together every setting yourself.
Like the rest of Talent Matching, the assistant is assistive AI. It doesn't evaluate, score, rank, advance, reject, or disposition candidates, and it doesn't make hiring decisions. It suggests a setup. Your team reviews it, applies it, and decides what to do next.
Turn on the Talent Matching assistant
The assistant uses the same infrastructure and permissions as Talent Matching, but it has its own AI toggle. A Site Admin needs to turn it on before anyone in your organization can use it.
Go to Configure > AI Tools. In the Real Talent section, confirm that Resume parsing and Talent Matching calibration are turned on. Both are required for full Talent Matching functionality, including match scores. Turn on the Talent Matching assistant toggle.
Note: Organizations on legacy subscription tiers can't use Talent Matching or the assistant. Both are limited to Real Talent customers on Core, Plus, and Pro.
For help setting up Talent Matching and Real Talent, see Talent Matching setup.
Open the assistant on a job
The assistant is available from the Talent Matching page on any job where Talent Matching is turned on.
Go to the job's dashboard and open the Talent Matching section. Go to the Application Review tab. Above the candidate table, click Configure with assistant.
The assistant opens as a panel over the candidate table. It has two columns:
- Suggested setup on the left, where the draft calibration and filters appear as you chat.
- The chat area on the right, headed Talent Matching assistant with a Beta tag, where you describe what you're looking for.
A footer at the bottom of the panel reads This tool uses AI. Please double check responses.
Note: The assistant opens with a short message telling you what it can help with. If the job already has a job post and scorecard attributes configured, it uses that context to draft an initial setup before you type anything.
Describe what you're looking for
The assistant is built so you can think in plain language instead of system settings.
In the Describe the candidates you want to find field, describe the role and the kind of talent you want to surface. For example, Mid-level account executives in New York with SaaS experience and at least 3 years in sales, or Senior product designers with a fintech background and strong research skills.
Send your message. The assistant shows a thinking state while it works.
The assistant interprets your description and drafts two things:
- A suggested calibration, which sets what matters most when scoring candidates for this role.
- A set of suggested filters, which are hard constraints on which candidates appear in the list.
You can keep chatting to refine the draft instead of starting over. Each follow-up message updates the Suggested setup column.
Tip: Paste a short summary from your job description or kickoff notes to give the assistant more to work with. The more specific your description, the more useful the draft.
Review and apply the suggested setup
Nothing is saved to the job until you apply it. Use the Suggested setup column to check the draft first.
- Suggested calibration shows a draft calibration built from your description and the job's existing context. It can include skills with importance levels, job titles, minimum years of experience, and preferences such as Include similar roles and Prioritize current experience.
- Suggested filters shows the draft filter set. The assistant uses the same filter types as the Talent Matching table, including location, keywords, application date range, and selected advanced filters.
Edit the calibration fields, then add or remove filters from the suggested list. When you're satisfied, click Apply setup.
When you apply the setup, the assistant does the following in order:
- Saves the suggested calibration as the active calibration for the job, if you have permission to manage calibrations.
- Applies the selected filters to the filter bar above the candidate table.
- Closes the panel and refreshes the table so you can review candidates against the updated match scores and filters.
To leave your setup as it was, click Cancel or close the panel. Closing the panel doesn't change your calibration or filters.
Understand what the assistant can and can't do
The assistant configures your search. It doesn't run the search for you, and it doesn't reach beyond Talent Matching setup.
It won't make hiring decisions. The assistant doesn't evaluate, score, rank, or disposition candidates. It suggests calibration and filters, and your team decides what to do with them.
It works one job at a time. The assistant is scoped to the job whose Talent Matching page you're on, and it doesn't carry context between jobs. You can't reopen an earlier assistant session; each time you open the assistant, you start a new session for that job.
It's desktop only. Like the rest of Talent Matching, the assistant isn't available on mobile.
Note: The assistant can run in filters-only mode. If AI match scoring is turned off for the job, or you don't have permission to manage calibrations, the assistant tells you so and suggests filters only.
AI guardrails and data handling
The Talent Matching assistant follows the same responsible AI pattern as Talent Matching and other Real Talent features.
Assistive, not automated
The assistant proposes a setup and waits for you. Calibration is never applied automatically. A person reviews and saves it every time.
Permissions and scope
The assistant respects your existing Greenhouse permissions and Talent Matching access. It can only draw on data you're already allowed to see. Access is gated by both Talent Matching and the Talent Matching assistant AI toggle.
Data processing
The assistant is built on the same underlying data and infrastructure as Talent Matching and may use third-party models to interpret your input and generate suggestions, under Greenhouse's existing AI terms.
For details on how Talent Matching processes data, including roles, retention, and opt-out, see: