The Workable Agent consumes AI credits as it works through your candidates — for example, when it screens them against your ideal candidate profile, sources passive candidates, or engages them. Because spend is driven by how much the Agent does, the main way to control AI credits is to manage how many candidates reach it and how much work it takes on. This article covers the practical levers: tightening your requirements and application form, using knock-out questions, unpublishing jobs, and turning the Agent on only where you need it.
How AI credits are consumed
The Agent consumes AI credits for each distinct action it takes on a candidate:
- Evaluate: 1 AI credit each time a candidate is screened against your ideal candidate profile.
- Source: 2 AI credits each time a passive candidate is identified and added to your pipeline.
- Chat: 10 AI credits each time a candidate interacts with the chat, meaning that they provide at least one answer to the chat even if they don't fully complete the chat session. If they open and close the chat without providing any input, then no AI credits are consumed.
These actions can stack on the same candidate, so the total AI credits used depends on how far that candidate progresses. A few cases use no AI credits at all:
- Candidates auto-disqualified by an application form knock-out question, because the Agent skips them entirely.
- Candidates you have blocked.
- Candidates created while the Agent is switched off on the job.
So the levers that actually reduce spend are the ones that reduce how many candidates reach the Agent and how much work it takes on.
Note: Disqualifying criteria in your ideal candidate profile are different from application form knock-out questions. A candidate disqualified by an ideal-candidate-profile criterion is still evaluated first, so it uses 1 AI credit. A candidate knocked out by an application form question is never evaluated, so it uses none.
Example consumption scenarios
Here's how AI credits add up in a few common cases:
- Applied candidate, no chat: the Agent evaluates the candidate against your ideal candidate profile. Total: 1 AI credit (evaluate).
- Applied candidate, completes chat: the Agent evaluates the candidate, then the candidate interacts with the chat. Total: 11 AI credits (1 evaluate + 10 chat).
- Sourced candidate, completes chat: the Agent sources the passive candidate, evaluates them, and then the candidate interacts with the chat. Total: 13 AI credits (1 evaluate + 2 source + 10 chat).
Reduce the number of candidates the Agent screens
Every applied candidate costs at least 1 AI credit to evaluate, so most AI credit use comes from applicant volume — meaning the biggest savings come before the Agent ever runs:
- Define clear requirements: an accurate job description attracts fewer, more relevant applicants
- Use a focused application form: a tighter application form discourages low-fit applicants from applying
- Add knock-out questions: set auto-disqualify application form questions for your true must-haves so unqualified applicants are knocked out before the Agent screens them.
Tip: Knock-out questions are the most direct way to cap spend on applied candidates. A candidate who fails a knock-out question is disqualified at the point of applying and uses no AI credit.
Unpublish a job once you have enough applicants
When a role has attracted a strong enough pool, change its status so new applicants stop entering the pipeline. See managing job statuses for how to unpublish a job.
Turn the Agent on only where you need it
The Agent only consumes AI credits on jobs where it is switched on. To keep spend predictable:
- Enable the Agent only for the jobs that genuinely need automated sourcing and screening.
- Switch the Agent off on a job once you already have a good talent pool, so no further candidates are processed there. Stopping the Agent for a job means that candidates already invited to chat will lose access unless they entered the chat before the stop, in which case they won’t be affected.
Candidates created while the Agent is off on a job are not processed and use no AI credits.
Manage autosourcing
When sourcing is enabled, the Sourcing Agent adds qualified passive candidates to the Sourced stage, and each passive candidate it adds costs 3 AI credits in total (1 to evaluate, 2 to source). The Sourcing Agent works toward a target of roughly 10 interested candidates per job, but it may need to reach out to many passive candidates to get there, so sourcing is not a fixed cap on AI credits. If you want to limit this, switch off the Source passive candidates action for the job once you have enough candidates in the pipeline.
Manage or switch off the Agent
You can stop the Agent on a single job at any time, for example while you adjust your setup or when a role is on hold. While the Agent is stopped on a job:
- All Agent operations halt, including screening candidates against your ideal candidate profile.
- Candidates created during this time are not processed, so they use no AI credits.
- Follow-up messages stop, and any chat links already sent to candidates stop working.
When you switch the Agent back on, it reprocesses your eligible recent candidates from the last 7 days. Older candidates from the stopped period are not reprocessed automatically, but you can start processing them on demand if you have AI credits available.
To stop all AI across the account, an admin can switch off AI features in settings. One control covers all of Workable's AI, including the Agent and scoring.
FAQs
- Can I set a daily or weekly limit on how many candidates the Agent contacts?
No. There is no daily or weekly contact cap. You control spend by reducing applicant volume (clear requirements, a focused application form, and knock-out questions), unpublishing jobs once you have enough candidates, and switching the Agent off on jobs that no longer need it.
- Can I set a limit on how many AI credits the Agent can use per job?
No. There is no per-job or per-account credit cap, and the Agent keeps processing candidates as long as account AI credits are available.
- If I stop the Agent mid-process, what happens to candidates already in a chat?
The Agent stops taking new actions and stops sending follow-ups. Chat links that were already sent stop working, so a candidate cannot resume a conversation once you have stopped the Agent for that candidate or job. Any AI credits already consumed for that candidate are not refunded.
- How do I stop the Agent completely?
Switch the Agent off on the job, or have an admin switch off AI features for the whole account in settings.