SleekAI reads your job posts - whether they live as WordPress posts, are embedded from your ATS, or are mirrored via API - and links to the ATS application page with the job pre-selected. It doesn't replace the ATS; it sits in front of it as the candidate-facing layer. The application, interview scheduling, and offer flow all stay in the system your team already uses.

Only if you've published them. Many companies now publish bands for legal compliance - California, Colorado, New York, several EU member states under the EU Pay Transparency Directive. SleekAI quotes those bands exactly as written. If you haven't published bands, the bot defers ('I can connect you with the recruiter for compensation details') rather than inventing numbers.

Yes - multibot lets engineering, design, sales, and operations each have their own assistant with department-specific tone, content scope, and starter questions. An engineering bot can talk stack and on-call; a sales bot can talk quota and territory. Candidates get more relevant answers and your hiring leads see questions sorted by team automatically.

It can capture qualifying info - years of experience, location, current role, eligibility to work - but actual hiring decisions should remain with humans. We recommend using the bot to inform recruiters rather than to filter automatically: hard filters in chatbots have a poor track record on bias, and the legal exposure isn't worth a marginal time saving.

Conversations live in your WordPress database under your control. Using your own OpenAI API key with the standard API terms means candidate data is not used for model training by default. Set retention policies to match your hiring-data policy - typically aligned with your ATS retention - and document the bot in your candidate privacy notice.

Yes - candidates writing in German, French, Spanish, Mandarin, or any common language get answers in that language, with role details translated on the fly. For international hiring this is a real lever - a candidate gets fluent answers in their first language about a role advertised in English, which materially shifts who applies.

Configure the system prompt to defer sensitive questions to a human recruiter rather than improvise - the model won't know your funding runway or current strategy, and a wrong answer is worse than a polite 'the recruiter can speak to that on the call'. Honesty about the bot's limits is more reassuring to a senior candidate than a confident guess.

We recommend not. Letting the bot write candidate-facing application material crosses a line that hiring teams generally regret - it muddies signal in the funnel, candidates assume bias if the same template appears repeatedly, and the legal posture is awkward. Keep the bot scoped to information delivery and let candidates write their own application.