Kula uses AI in more places than just Kula Teammates. This article is a single inventory of every AI-powered feature in the product — what each one does, whether it acts on its own or just informs a person, whether it costs extra, and what candidates are told about it. Use it as a quick reference for internal review, security questionnaires, or explaining your AI footprint to a candidate or customer.

AI features that inform a person (no autonomous action)

These generate content or analysis for a recruiter, hiring manager, or candidate to read and act on. They don't reject, advance, or schedule anyone on their own.

• AI Application Assistant. On a candidate's profile, generates and keeps refreshed a summary of the candidate, and answers free-text questions about them with citations back to source data — scorecards, emails, notes, transcripts, application details.

  Respects existing permission rules (private notes only go to the people they're already shared with; restricted application fields and offer details stay restricted) and its scope is limited to the current job — it won't surface cross-job detail beyond the bare fact that a candidate applied elsewhere. Kula states this explicitly: the Assistant summarizes information but does not make hiring decisions or recommend whether to hire or reject.

• AI Notetaker. Joins Google Meet/Teams interviews, transcribes, and can autofill scorecards from the transcript. Included in the standard plan (confirmed for at least some accounts via a direct CSM explanation to a prospect). Doesn't yet support phone-call interviews, and in-person capture is unvalidated. See "Kula AI overview" and the AI Notetaker setup articles.

• Resume parsing and autofill. Reads an uploaded resume (PDF/DOCX most reliably) and pre-fills parts of the application form or candidate profile. Coverage is partial today — see "The candidate-facing application experience" for exactly which fields it reliably fills.

• AI-Generated Messages (Dynamic AI Block). In Flows/outreach sequences, generates industry- and role-tailored messaging content for candidate outreach. See "What is the Dynamic AI Block in AI-Generated Messages?" and "Best Practices for Using AI-Generated Messages and the Dynamic AI Block."

• AI job description generation ("Ask AI"). An Ask AI button when creating a job post drafts a job description for you to edit. See "How to Create a Job Description."

• AI Candidate Search. Lets a recruiter search the candidate database in natural language (for example, "candidates from Chennai with SaaS background") instead of building manual filters. Shipped. Its companion feature, AI-recommended candidates surfaced proactively rather than searched for, has not been built yet — don't describe it as available.

• Conversational Analytics. A natural-language chat interface over your hiring reports and data — ask a question in plain English instead of building a report manually.

AI features that can take action

These can score, flag, or act on a candidate — either automatically or once you've explicitly configured them to.

• Screener (AI Scoring). Screener and "AI Scoring" are the same feature under two names. It scores candidates against job criteria. On its own it informs a recruiter's decision, but it can be configured to drive automatic rejection (via Knockout Questions or auto-reject rules built on its output) — see "Knockout questions" for exactly how that's scoped. Included in the standard plan; no org-level settings page, configured per job.

• Fraud detector. Assesses a candidate's resume, LinkedIn profile, email, phone, location, and device information — cross-referenced against public information and application patterns — and returns a verdict: No issues, Inconclusive, or Suspicious. It does not auto-reject unless you've explicitly configured that. Paid add-on, behind a feature flag that Kula enables per contract. See "Fraud Detector."

• Coordinator. An autonomous phone-scheduling agent — it calls candidates and books interviews on its own once activated. Every call is recorded and transcribed. Requires purchasing phone numbers. Paid add-on.

What is not AI

• Knockout Questions / auto-rejection rules are deterministic, rule-based logic (if a custom field answer matches a configured condition, reject) — not AI or machine learning, and evaluated the same way every time regardless of Screener being on or off.

• It's easy to conflate the two because both can result in an automatic rejection; see "Knockout questions" for the mechanics.

• If you're filling out a security or AI-disclosure questionnaire, Knockout Questions don't belong in the AI section.

AI disclosure and candidate consent

• Kula doesn't currently have one account-wide AI-disclosure toggle. Today, the standard approach is to disclose AI use in the job description and/or add a Yes/No consent question as a custom application form section (Job → Job Post → Application Form → Add Section → Add Question).

• Fraud detector is the one exception that's automatic: when it's enabled on a job, a disclosure that the organization uses AI to screen for fraud is added to that job's application page without any extra setup.

• For broader AI-compliance questions (GDPR, EU AI Act, and similar), Kula publishes an independent third-party audit at trust.warden-ai.com/kula/ai-scoring, including an EU AI Act–specific page, refreshed via monthly audits — this is the standard first response to a customer's compliance or security-review question about Kula's AI.

• It has been the deciding factor in at least one real enterprise deal's compliance review.

• It does not, on the material reviewed for this article, confirm coverage of every state-specific AI-hiring law (see Needs verification for the Illinois example) — for a law not covered on that page, escalate rather than assume it's covered.

Good to know

• Only two features require a paid add-on unlock: Fraud detector and Coordinator. Screener/AI Scoring and AI Notetaker are included in the standard plan for at least some accounts.

• Every Kula Teammate (Fraud detector, Screener, Coordinator) needs two independent switches on: the org-level toggle in Settings → AI & Kula teammates, and that specific job's AI Teammates tab. Turning it on org-wide does not activate it on any job by itself.

• Kula does not train or fine-tune AI models on customer data — confirmed for AI Notetaker's interview recordings, transcripts, and summaries specifically, which are treated as your data under the DPA.

• AI Notetaker has known, real accuracy limitations worth setting expectations around: summaries can lean too positive and miss red flags called out verbally in the interview; the model can occasionally state something not actually in the transcript (misnaming, stale assumptions); and scorecard autofill sometimes completes only partially (for example, filling the summary but skipping other fields). More structured, competency-based scorecards produce more reliable AI-filled feedback than a single general feedback field.

• Conversational Analytics' totals currently don't reliably match the Reports numbers — an open, actively tracked issue. Don't cite CA output as a final number in an external report until this is resolved.

• The AI Application Assistant is explicitly scoped to summarize and answer questions — not to recommend a hiring decision. Screener/AI Scoring is different: it can be configured (via Knockout Questions or auto-reject rules) to actually reject a candidate automatically. Know which category a given AI feature falls into before describing it to a candidate or auditor.

FAQ

• Do I need to tell candidates we use AI in hiring? Increasingly this is a legal requirement in some jurisdictions, and it's good practice everywhere. Today, add disclosure to the job description and/or a consent question on the application form. Fraud detector adds its own disclosure automatically when enabled.

• Does Kula train its AI models on our data? Confirmed no for AI Notetaker's interview content — it's treated as your data under Kula's DPA, not used to train models.

• Is Kula's AI scoring compliant with regulations like the EU AI Act? Kula maintains an independent third-party audit dashboard covering this, including an EU AI Act–specific page, refreshed monthly: trust.warden-ai.com/kula/ai-scoring. For a specific state or national law not addressed there, check with your account team rather than assuming coverage.

• Which AI features cost extra? Fraud detector and Coordinator are paid add-ons. Screener/AI Scoring and AI Notetaker are included in the standard plan for at least some accounts — confirm your specific contract for certainty.

• Are Knockout Questions an AI feature? No — they're deterministic rule-based logic, evaluated the same way regardless of any AI feature's settings. See "Knockout questions."

• Can any AI feature reject a candidate without a person reviewing it first? Only if you've explicitly configured it to — for example, Screener/AI Scoring feeding a Knockout Questions rule, or Fraud detector configured to auto-reject on a Suspicious verdict. Neither does this by default, and the AI Application Assistant never recommends a hiring decision at all.

Need help?

If you have questions about a specific AI feature, its data handling, or compliance documentation, reach out to us at support@kula.ai or use the in-app chat.