AI Features in candidate.fyi
Last updated: August 21, 2026
Approximate read time: 12 minutes
This article describes the AI features currently available in candidate.fyi. It is written for customer administrators, legal teams, and privacy reviewers who need to understand what each feature does, who uses it, how it is enabled, and where human oversight applies.
AI features in candidate.fyi support recruiting operations such as interview scheduling, coordination, note-taking, and reporting. They are not used to score, rank, or select candidates. Your organization remains responsible for all employment decisions.
How AI features are enabled
Most AI features are optional and must be turned on for your organization before they appear in the product. candidate.fyi enables features at the organization level through internal configuration (often called feature flags). If a feature is not enabled for your organization, the related screens, toggles, and menu items will not appear.
Overview of AI features
Feature | Primary users | Enabled by default? |
|---|---|---|
Insights Ask / Ask FYI | Internal (editors and above) | No — organization must enable |
Interview Companion / AI notetaker | Internal (hiring team); candidates see consent | No — organization must enable; optional per interview |
MCP connector | Internal users authorizing external AI tools | Available when users connect via OAuth |
AI Availability input | Internal schedulers | Appears available when users can enter availability (no separate org toggle found) |
Scheduling request AI summaries | Internal schedulers | No — organization must enable |
Coordination agent | Internal schedulers | No — organization must enable; assigned per request |
Scheduling follow-up AI assistant (Nudge) | Candidates receive emails; internal users configure | No — organization must enable |
Insights Ask / Ask FYI
What it does
Ask FYI lets internal users ask natural-language questions about recruiting and scheduling data — for example, how many interviews were scheduled in a given period, or trends across interviewers or departments.
There are two related experiences:
Insights > Ask — a chat-style interface where you can choose Ask (answer questions about your data) or Build (create and edit a metric). Placeholder text includes “Ask fyi anything about your operations.”
Insights > Reports > AI Query — a search-style interface with placeholder text such as “Ask about your analytics...” and an Ask Question button. A footer note states: “Results are AI-generated and may contain errors. Verify critical details before acting.”
Who uses it
Internal users only — typically editors and administrators with access to Insights. Candidates do not use this feature.
How it interacts with AI models
Your question and relevant operational data from your organization are sent to AI models to produce an answer or suggested metric. This data only used for inference (to generate a response). The product displays a disclaimer that results may contain errors.
Human oversight vs automation
Ask FYI is read-only. It does not change schedules, send emails, or update records. Users must review answers and decide how to act.
Consent, notice, and retention
No separate consent flow or retention settings are shown for this feature beyond the accuracy disclaimer on AI Query.
Interview Companion and AI notetaker
What it does
When enabled and assigned to an interview, the Interview Companion can:
Join interviews (including via dial-in for phone interviews)
Record and/or transcribe the conversation (depending on configuration)
Generate an Interview Summary, Interview Transcript, Q&A highlights, and Timing insights
Support Ask fyi questions grounded in a specific interview’s transcript and recording
Optionally autofill interview feedback forms (when that sub-feature is enabled)
On the interview detail page, generated content is labeled fyi Generated. Users can open How fyi Analyzes Interviews to see how analysis is produced.
Who uses it
Internal users (recruiters, coordinators, interviewers) view summaries, transcripts, and insights.
Candidates see a Recording & AI notes card during scheduling flows. They can choose I consent or I do not consent. Status badges include Recording: On, Recording: Off, and Consent required.
The feature is optional. It must be enabled for the organization, configured under Settings > fyi > fyi Agents > Interview Companion, and may be applied per interview, per scheduling request, via booking links, or through personal defaults on Profile > Interview Companion.
How it interacts with AI models
Interview audio and/or transcript content is processed to produce summaries, Q&A highlights, timing insights, and responses to Ask fyi questions about that interview. The AI models analyze responses for technical depth, communication clarity, role fit, and related signals.
Human oversight vs automation
Recording may start automatically when configured (depending on consent model). Generated notes are intended to support — not replace — human review. The product states: “fyi provides analysis to support your decision-making, but human judgment remains paramount. Always review the evidence and add your own perspective.”
Consent, notice, and retention
Administrators configure consent and retention under Settings > fyi > fyi Agents > Interview Companion:
Capture mode: Full capture (audio/video + transcript) or Transcript only
Candidate consent model: Opt-out (recording enabled by default) or Opt-in required
Consent message — customizable message shown to candidates, with an optional Recording Consent Link token
Media & transcript retention — Indefinite, 7 Days, 14 Days, 30 Days, 60 Days, 90 Days, 180 Days, or One Year
Notes retention — same options as media retention
Default candidate-facing copy (when no custom message is set) explains that AI notes help interviewers focus on the conversation, that the recording is used to generate a summary for the hiring team, that content is only shared with the hiring team, and that candidates can change their choice via their confirmation link.
MCP connector
What it does
The MCP (Model Context Protocol) connector lets external AI assistants — such as Claude — connect to candidate.fyi on a user’s behalf. After authorization, the assistant can perform actions such as searching and scheduling interviews, finding availability, listing calendars, updating interviews, and (when reporting features are enabled) querying analytics.
Who uses it
Internal users only. When connecting, the user sees an OAuth screen: “{Application name} would like access to your account,” chooses an organization, and clicks Allow access. Candidates do not use MCP.
How it interacts with AI models
MCP connects an external AI assistant to your candidate.fyi data. What that external assistant sends to its own AI models is controlled by the third-party application, not candidate.fyi.
Needs confirmation from the team: Whether candidate.fyi has a separate organization-level toggle to enable or disable MCP beyond OAuth authorization.
Human oversight vs automation
Access requires explicit user authorization. Scheduling changes require appropriate internal permissions (typically editor-level access or above).
Consent, notice, and retention
The OAuth consent screen is the primary notice. No retention settings are shown on the consent screen.
AI Availability input
What it does
When entering candidate availability, internal users can describe availability in plain language instead of selecting each slot manually. For example, in the Candidate Availability modal, the Ask AI field accepts text such as “every day next week from 12–3pm,” and the Generate button converts it into structured time slots. Parsed slots may show an AI badge. Users can review, edit, or remove slots before saving.
A similar input appears in the Chrome extension under Candidate availability with helper text: “Describe the candidate’s availability or paste their message.”
Who uses it
Internal schedulers and recruiters only. Candidates use structured date/time pickers in the candidate portal, not this AI input.
How it interacts with AI models
The natural-language text you enter is sent to AI models to propose structured availability slots.
Human oversight vs automation
AI proposes slots; a human must review and confirm before availability is saved.
Consent, notice, and retention
No separate consent or retention settings are shown for this feature.
Scheduling request AI summaries
What it does
On a scheduling request’s Interviews tab, a fyi AI Summary card provides an AI-generated overview of the request context — drawn from notes, emails, and related activity. The card shows when it was last updated and includes a refresh control (Refresh AI summary).
Who uses it
Internal schedulers and recruiters. Not visible to candidates.
How it interacts with AI models
Scheduling request content (notes, emails, threads, and related context) is sent to AI models to produce the summary.
Needs confirmation from the team: Whether content is used only for inference or also for model training.
Human oversight vs automation
Summaries are generated automatically and on refresh. They are informational — users decide how to act on them.
Consent, notice, and retention
Enable or disable under Settings > fyi > fyi Settings > Summarization > AI scheduling request summary. No retention settings are shown.
Coordination agent
What it does
The Coordination agent helps align availability, handoffs, and follow-ups across candidates, interviewers, and schedulers. When assigned to a scheduling request, it can work across email threads to support scenarios such as availability submission, reschedule requests, candidate withdrawal, and interview support requests.
On an assigned request, users may see tabs including Plan, Conversations, Agent Work, and Escalations, plus a chat sidebar with prompts such as “How can fyi help with this request?”
Who uses it
Internal schedulers and coordinators. The agent may read and respond on email threads involving candidates and interviewers when autonomy settings allow. Candidates interact with the agent indirectly through email, not through a dedicated agent UI.
How it interacts with AI models
Email content and scheduling context are processed to propose or carry out coordination actions.
Human oversight vs automation
Administrators configure Agent autonomy under Settings > fyi > fyi Agents > Coordination Agent:
Shadow mode — the agent proposes actions for human review
Autonomous mode — the agent can act directly
Individual capabilities (Interview Support Request, Candidate Withdrawal, Availability Submission, Reschedule Request) can each be set to disabled, shadow, or autonomous. The Escalations tab surfaces items that need human attention.
Consent, notice, and retention
No separate consent or retention settings are shown in the Coordination agent configuration.
Scheduling follow-up AI assistant (Nudge)
This capability is configured through the Scheduling Agent (under Settings > fyi > fyi Agents > Scheduling Agent), not the Coordination agent.
What it does
When enabled, fyi can automatically send follow-up emails to candidates and reminders to interviewers. Configurable options include:
Scheduler Reminders
Candidate Followup Nudges — follow up with candidates to enter availability or schedule their interview
Interviewer Attendance Nudges
Candidate Interview Reminders
Auto-close on followup maxout — close the request if there is no response within 48 hours of the last nudge
Auto-close on candidate opt-out — detect withdrawal intent in email replies and close the request
On individual scheduling requests, an AI Followup Agent toggle (under request settings) controls whether follow-ups run for that request.
Candidate-facing emails identify the sender as an AI assistant, for example: “{Name}” signed as AI Assistant supporting {Recruiter first name}.
Who uses it
Candidates receive automated follow-up and reminder emails.
Interviewers may receive attendance nudges.
Internal users configure the Scheduling Agent and per-request toggles.
How it interacts with AI models
Outbound follow-up emails use predefined templates with dynamic fields (candidate name, scheduling links, etc.). AI models will never create or write emails. Inbound candidate email replies may be classified by AI — for example, to detect disinterest in the role, a reschedule request, or a message that requires a human response.
Human oversight vs automation
Follow-ups send automatically according to agent and request settings. When a reply is classified as requiring a human response, or when auto-close on opt-out is enabled, the system routes toward human handling rather than continuing automated follow-ups.
Consent, notice, and retention
Email signatures disclose that the message comes from an AI assistant supporting a named recruiter. No separate privacy page or retention setting was found for follow-up emails.
AI bias and employment decisions
candidate.fyi’s AI features are designed to support recruiting operations — scheduling, coordination, note-taking, and reporting — not to evaluate or compare candidates for hiring decisions.
AI features do not score, rank, or select candidates.
Interview Companion analysis highlights skills, communication, role fit, and potential risks to help interviewers review evidence — but the product explicitly states that human judgment remains paramount.
Your organization is responsible for all employment decisions.
candidate.fyi does not claim fairness audits, bias certifications, or similar third-party evaluations unless separately documented by your account team.
Model training and data use
Across the features above, candidate.fyi sends relevant text, scheduling data, interview recordings/transcripts, and email content to AI models to generate responses, summaries, classifications, or structured data.