AI that takes the action, not just the notes
The AI Recruiter works in your pipeline rather than describing it. Thirteen tools let it look records up, create jobs and candidates, link submissions, schedule interviews, send email and search connected job boards — every call checked against the same permissions you already have.
Where AI shows up in Jwalix
Each one attaches to a module you already use — no separate AI product to open, and nothing that needs your data exported somewhere else first.
AI Recruiter
An assistant that calls real actions in the platform — creating records, scheduling, emailing and reporting — instead of handing you text to copy somewhere else.
AI interviews, by link or phone
Structured first-round screening a candidate completes on their own schedule, either through a public link or as an inbound phone call.
Semantic matching
Surfaces candidates already in your database whose experience matches a requisition in meaning, not just in shared keywords.
MCP integration
Point Claude or your own agent at your workspace and let it run the same tools, under the same permission checks, through a revocable API key.
What the AI Recruiter can actually do
Thirteen defined actions, each with a validated set of arguments. Asking in plain language picks the tool; it does not invent a new one.
Look records up
Find an existing candidate by name, email, code or id; a job by title or code; a submission by code — then keep working with what it found.
Create records
Draft a new job, add a candidate, or link a candidate to a job as a submission. These run only when you explicitly ask for something new to be created.
Schedule interviews
Book an interview against an existing submission, so it lands on the same record as everything else.
Send email
Send to a candidate, a contact, or an address you name, without leaving the conversation.
Search for talent
Search your own database and connected job boards — Dice, Foundit, Naukri and LinkedIn — from one request instead of four tabs.
Answer with your data
Counts and summaries across candidates, jobs and submissions, filtered to a timeframe or grouped by owner, drawn live rather than from a stale export.
Build a recruiter brief
Assemble a candidate fit report, job brief, submission brief, interview kit or outreach draft in one step.
Decide what to work on
Ask it to review the pipeline and recommend the highest-value actions available right now, rather than deciding from a list you have to read first.
Apply what is new
Summarize where AI recruiting practice is heading and, more usefully, which parts of it your workspace is already set up to use.
An assistant, on your permissions
The interesting question about an AI that can create records and send email is what stops it. Three things do.
It runs as you
Every tool call goes through the same module-action permission checks as the equivalent click. If you cannot delete a candidate, neither can the assistant on your behalf.
Creating is explicit
The create actions are defined to fire only on a clear request to create something. Asking a question about a job does not quietly produce a new one.
One definition, everywhere
The in-app assistant and the MCP surface read the same tool definitions, so a permission or argument change applies to both at once and they cannot drift apart.
Worth being clear about
AI output is a starting point, not an approval step. Drafted email is still yours to read before it sends, a created record is still a draft you can correct, and interview results are evidence for a human decision rather than the decision itself.
Questions about AI in Jwalix
No. It is reachable from the same navigation you already use, and works against the record you are looking at rather than a copy of your data kept somewhere else.
It can do both — create jobs, candidates and submissions, schedule interviews and send email. Those actions run under your own permissions, and the create actions only fire when you explicitly ask for something new.
Nothing about being an AI request widens access. Each tool call is checked against the same module-action permissions that gate the equivalent action in the interface.
No. It is used for structured first-round screening. Candidates who pass move into scheduled interviews with your team, tracked on the same application record.
Yes. The MCP integration exposes the same thirteen tools to any MCP-aware client, authenticated by a revocable API key that carries your permissions rather than bypassing them.
See AI at work across the platform
Watch the AI Recruiter work a real pipeline
We'll run the assistant, an AI interview and matching against your own candidate data on the call.