You applied through a careers page, closed the tab, and a few minutes later an email arrived inviting you to chat with an AI recruiting assistant. At HR Tech 2026 in Las Vegas, Workable shared what happens next: of more than 350,000 candidates its agent contacted, 37% never opened the conversation. Most of the people who drop out do it in their inbox. Very few drop out because of anything the bot asks.
What is the Workable AI agent message you get after applying?
It's an email with a link to a text chat run by the Workable Agent, an AI recruiting assistant built into Workable's applicant tracking system. It isn't an SMS. Workable's help docs say "a candidate receives an email to join the chat session," and the agent always tells you it's an AI. Employers can't turn that disclosure off.
The agent launched in May 2026. Since then more than 1,700 Workable customers have used it on nearly 12,000 jobs. CEO Nikos Moraitakis describes it as "a true teammate inside your ATS," and in practice it does three jobs:
- Scores you. Each candidate gets a score out of 100 against the employer's Ideal Candidate Profile. That profile is the list of criteria the hiring team says the right person meets.
- Fills in blanks. Each criterion is marked met, partially met, not met, or "unknown." If something is missing from your application, the agent asks targeted follow-up questions and then "re-scores the candidate instantly," according to Workable.
- Hands off a transcript. Recruiters get a detailed evaluation and the full conversation log.
Hard knockout questions are separate from the chat. A knockout question is a pass/fail item such as work authorization or location, and employers put those on the application form itself. So if you're in the chat, you have most likely already cleared them.
Why does getting the message matter?
Getting the chat invite is a good sign: Workable's agent evaluated 1.2 million applications and flagged only about 350,000 for outreach, so roughly 29% of applicants got a message at all.
The mechanism explains why it matters. The agent reaches out to resolve "unknown" criteria so it can re-score you. Workable doesn't say how an unresolved unknown is scored. If you never open the chat, though, the blank stays blank, and a reviewer comparing you with someone whose criteria are all resolved has less reason to call you. That last part is my inference, not a published rule. The safe move is to fill in the blanks yourself.
The other 71% of applications got nothing back. If you send dozens of applications a week, that's how most of them feel: a form goes out, an automated system reads it, and you never learn which criterion you missed. That silence is the real tax of high-volume applying, and responding faster to the chats you do get only fixes part of it.
Where do candidates drop out of the Workable funnel?
Most of the loss happens in the inbox: 37% of flagged candidates never opened the chat. Here is how Workable's HR Tech 2026 numbers break down when you work them through.
| Stage | Count | Share of flagged |
|---|---|---|
| Applications evaluated | 1,200,000 | 29% flagged |
| Flagged for outreach | ~350,000 | 100% |
| Opened the chat | ~220,500 | 63% |
| Never opened | ~129,500 | 37% |
| Completed the chat | ~169,800 | ~48.5% |
| Asked for a human | 222 | 0.06% |
These are rough figures. Workable said "more than 350,000" were flagged and that "nearly 200,000" moved on to recruiters or hiring managers, which is more than the roughly 170,000 completers you get from 63% times 77%. Read the table as the shape of the funnel, not as exact accounting.
The shape is still clear. Two drop-offs matter, and you control both:
- Not opening. About 37% of flagged candidates never clicked.
- Not finishing. 23% of the people who opened the chat didn't complete it. That's roughly 50,000 people who got an invite and stopped partway through.
They had already been flagged for outreach, from about 29% of 1.2 million applications.
The causes are probably ordinary. The email looks automated, or it arrives while you're at your current job, and you plan to come back to it. Then the deadline catches you.
How long do you have to respond to the AI recruiter chat?
You have 6 days to open the link and 7 days to finish once you start. Workable's help docs say an unopened link expires after 6 days however many reminders went out, and reminders arrive every 2 days.
| Window | Limit | What happens after |
|---|---|---|
| Outreach after applying | Typically within minutes | Email with chat link |
| Reminders | Every 2 days | They don't extend the deadline |
| Opening the link | 6 days | Link expires |
| Finishing once opened | 7 days | Status shows "expired" |
The "expired" status is the one to avoid. It applies to chats that were started and never finished, so a half-done conversation doesn't quietly disappear - it's logged as unfinished.
So treat it like a short phone screen you scheduled:
- Open it the same day. Hiring can move fast on Workable. Workable says its own three-person recruiting team filled a Head of Operations role in 7 days from posting to offer. CacheFly, a Workable customer, processed 930 applications in 36 hours.
- Finish it in one sitting. The agent "may continue asking follow-up questions until all required information is collected," so set aside time for a few rounds of questions.
- Search your inbox and spam for the company name and "Workable" every day for a week after each application.
How to answer AI screening questions so they match your resume
Your chat answers have to say the same thing as your resume, because the recruiter reads them side by side. Recruiters get the full transcript next to your application. If your chat gives a different start date, title, or tool than your resume, that mismatch is now on record.
Consistency counts for more than charm. This is a text chat, and the agent wants specific facts. A nice-sounding answer that contradicts your own resume is the worst outcome, and it usually happens because you typed from memory against a resume you tailored three weeks ago and no longer remember.
Build a fact sheet before you open the chat
A fact sheet is a single page of the answers you'd otherwise improvise. Keep it open in another window:
- Dates. Start and end month and year for every role on the resume you actually submitted.
- Titles. Exactly as written on that version, including any rewording you did for this posting.
- Location and work authorization. Where you live, whether you'll relocate, and whether you need sponsorship.
- Salary range. One range you're comfortable defending, consistent across applications to the same company.
- Notice period. How many weeks before you can start.
- Gap explanation. Two sentences, written out in advance (see below).
- Two questions for the hiring team. The agent asks for these before it closes.
The hard part is keeping a dozen tailored resumes straight. Refolk rewrites your resume from your own history for each posting, so the dates and titles stay the same across every version, and it keeps a record of each one, so you can pull up exactly what went to this employer before you open the chat.
Basic-fact questions
Answer basic-fact questions in one line with the number first. "Available in 4 weeks." "Based in [city], authorized to work in [country], no sponsorship needed." Don't hedge or explain unless the agent asks. Every extra clause is something a reviewer might read against your resume.
If a question touches a criterion you only partly meet, say what you have, specifically. "Two years of [tool] in production, mostly on [specific work]; no admin experience" gives the agent something to mark partially met. "Some exposure" gives it nothing to score, and the criterion may stay unknown.
AI screening follow-up questions about an employment gap
Workable says the agent can "clarify gaps in experience," so if your resume has one, expect a question about it. Answer in two sentences: what happened, then what you've been doing since, ending in the present.
A template:
I left [Company] in [month year] when [the team was cut / I relocated / I cared for a family member]. Since then I've [done contract work for X / completed Y / built Z], and I'm looking for a full-time [role] now.
That works for three reasons. It gives the reason without apologizing for it. It's short, so a follow-up has little to probe. And it puts a recent activity on record, so the reviewer reads the last line as current. If the agent asks a follow-up anyway, answer with a date or a deliverable. Don't add more explanation.
Write this once and keep it on your fact sheet.
Prepare the chat like a phone screen. The bot reads your answers once, but the recruiter reads them next to your resume.
Should I ask to talk to a human recruiter?
Usually not. Every candidate was offered a human handoff, and only 222 out of roughly 220,000 who opened the chat took it. Workable records the request as an explicit flag for the recruiter. But the chat itself feeds the recruiter's shortlist, so skipping it doesn't move you ahead.
That's about 0.1% of the people who opened the chat.
Sarfraz Ali, Workable's CRO, was careful about what that number means: "This is not saying candidates prefer talking to a machine." His point was that what candidates got before the agent "wasn't a human conversation either."
Ask for a human when the bot can't represent you accurately:
- You need an accommodation to complete the process.
- Your situation doesn't fit structured questions. Examples: a visa transfer in progress, a non-standard notice period tied to a retention bonus, or a role that spans two of the employer's criteria.
- The job information is wrong. The location, salary band, or level in the chat doesn't match the posting.
For anything else, use the "questions for the hiring team" step at the end of the chat. It reaches the same people, and you still finish the conversation that gets you scored.
Why walking away from the chat is the wrong protest
Many candidates don't trust AI screening, but refusing a text chat mostly hands your spot to someone else. In Greenhouse's 2026 Candidate AI Interview Report, which surveyed 2,950 job seekers in five countries, 38% said they had left a hiring process because it included an AI interview, and another 12% said they would.
The distrust has a basis. Gartner's survey of 2,918 candidates found only 26% trust AI to evaluate them fairly, while 52% believe AI already screens their applications. The second group is right, and with Workable that screening happens before the chat. When you get the invite, the AI has already read your application. The chat is your one chance to correct what it got wrong.
This is also a typed chat, not a video interview. There's no camera, and you can check your fact sheet while you answer. If you were going to quit over a recorded video screen, that's a different decision. Skipping this chat gives up something 63% of other candidates took.
FAQ
Is the Workable AI agent message a text or an email?
It's an email with a link to a text-based chat. Workable's help documentation says candidates receive an email to join the chat session. It can look like an automated confirmation and may land in a promotions or spam folder, so search your inbox for the company name for several days after you apply.
What happens if I miss the 6-day window?
The link expires and the conversation won't happen, no matter how many reminders were sent. Workable doesn't publish how an unanswered chat affects your score, but any criteria the agent needed to clarify stay unresolved. If you find an expired invite, reply to the employer through the careers page or another contact and ask whether the chat can be reissued.
Can I fix an answer I already gave the AI recruiter?
There's no edit button, and recruiters see the full transcript. If you notice a wrong date or title mid-chat, correct it in your next reply: "Correction: I started at [Company] in June, not May." If you only notice after the chat ends, add the correction to the questions you send to the hiring team, or raise it in the first human call.
Does asking for a human recruiter hurt my chances?
Nothing Workable has published says it does. Workable says the request shows up as an explicit flag for the recruiter. The risk comes from not finishing, because the completed chat is what feeds the recruiter's shortlist. Ask for a human when you need an accommodation or the bot can't capture your situation. For general questions, finish the chat and use its closing step.