The Wiggli Sourcing Agent

PUBLIC · Feature Overview

Why it matters

Search in Wiggli Intelligence already starts from a prompt instead of a Boolean string. The Sourcing Agent takes the next step. Instead of writing a prompt, adjusting filters, reading through a longlist, and pulling the strong profiles into a project by hand, you have a conversation. The agent asks the questions a sharp intake call would ask, runs the search, scores every profile against what you told it, and drops a ready shortlist into a project.

The work doesn’t disappear, it moves. You spend your time defining the role precisely, not operating the search.

How it works

The Sourcing Agent isn’t a single model. It’s a set of agents that hand off to each other, and you can follow each step live in the panel on the right.

1. Intake

The intake agent interviews you about the role. It asks follow-up and challenge questions: region and radius, whether the candidate must be in the role today or can be up to two years out, software or tooling requirements, the type of work (small clients and SMEs versus complex groups), language, minimum diploma, and what separates a strong candidate from an average one. Where it can, it offers answers as clickable options so you confirm instead of retype.

2. Sourcing

The sourcing agent launches the search. It maps which job titles exist in the region, works out which titles people in this role actually use (the same role gets called dossierbeheerder, accountant, or boekhouder depending on the person), builds and corrects the query, and pulls the top of the pool.

3. Screening and scoring

A screening agent first trims the longlist against your criteria, resolving synonyms along the way (GA4 and Google Analytics read as the same skill). The scoring agent then rates each remaining profile and explains the reasoning, why a profile is a strong match and why another is weaker, so you can audit the call rather than trust it blind.

What happens after the shortlist

The shortlist lands in a project, which is where the rest of your workflow already lives. From there you can run the AI Co-pilot to screen further with custom Intelligence Columns, reveal contact details, and push the shortlist into an outreach campaign or to your ATS. The Sourcing Agent replaces the manual front end of sourcing. It hands off to the same project you would have built yourself.

Getting the best results

Brief the intake like a real role. The shortlist is only as sharp as the intake. Answer the challenge questions properly, the agent uses them to for example: separate a dossierbeheerder from an accountant, not just to fill fields.
Run one session per region. The agent ranks the strongest profiles across the whole pool it searches. Search two regions in one session and the shortlist tilts toward whichever market has more supply. Run a separate session per location to get a clean, comparable shortlist for each.
Refine instead of starting over. Under the project title, Refine with Agent reopens the conversation with three options: refine and replace the candidates in this project, refine into a new project, or add more profiles like these (if only 25 more are relevant, it adds 25 and tells you so). Each run supports up to three refinement passes, so make each one count; for a real change of direction, start a fresh conversation.
Anchor on your best profile. Select a strong profile in the project and Discuss with agent appears. Ask for more profiles like that one and tell it which traits matter; the agent carries those criteria into the next search.
Say the hard limits out loud. Exclusions and caps (maximum 15 years of total experience, no freelancers, never these companies) only bind when you state them explicitly. The agent presents its brief before it starts searching: read it and correct anything missing before the run begins.
Talk first, upload second. You can hand the intake a job description, but the conversation is the point: it digs into what actually separates a strong candidate from an average one. If you start from a JD, make sure it’s well written and truly reflects the profile you want, because the agent takes it at its word.
Hand precise judgment calls to the Co-pilot. Criteria a ‘junior recruiter’ can’t fully resolve (must have worked inside an accountancy firm, must hold a specific degree) are well suited to an Intelligence Column in the project, where the Co-pilot evaluates each profile and tells you why.

Common misconceptions

It replaces the manual flow, not your judgment. The agent automates the search-to-project steps. You still set the brief and make the final call on the shortlist.
The shortlist is a starting point, not a verdict. If a profile scores strong but sits slightly off (a near-match job title, for instance), refine the conversation and the agent re-runs against the tighter brief.
It builds the project for you. You don’t create a project first. The agent creates one and places the scored shortlist inside it automatically.
It speaks your language. The agent responds in the language you type in: Dutch intake in, Dutch questions back.
It sees your projects, not your campaigns. The agent can exclude profiles already sitting in a project. It can’t yet check your campaigns, so scan a shortlist against running campaigns before enrolling it.

Still stuck?

Your customer success manager reads faster than any search bar.

Get help