Every AI agent in HR, mapped. And who decides.

12 areas · 122 agents

1. Workforce Planning & Org Design

   9 agents

   Workforce Planning & Org Design forecasts the headcount, skills, cost and structure the business will need, and shapes the organization to get there.

   3 proven · 2 strong case · 3 emerging · 1 human decides

   JobsForecast headcount and people cost against the business plan · Model what-if scenarios for growth, restructure or budget cuts · Understand skills supply versus demand · +4 more

   • Headcount Forecasting Agent Proven headcount and cost variance to plan; planning-cycle hours reclaimed

     What it does

     Pulls the approved hiring plan, historical attrition, open reqs and budget from HRIS and finance, and projects headcount and fully loaded people cost by team and quarter, refreshing as reqs open, close or slip.

     Who decides

     Finance and HR leadership set and change the plan; the agent projects, reconciles and explains variance, and never adds or removes a position on its own.

     Metric

     headcount and cost variance to plan; planning-cycle hours reclaimed

     Data it needs

     HRIS headcount and position data, hiring plan and open reqs, attrition history, finance budget and actuals

   • Workforce Scenario Modeling Agent Strong case planning-cycle time; quality of budget decisions (fewer mid-year re-plans)

     What it does

     Takes a scenario described in plain language (a hiring freeze in one region, doubling a sales team, a shift to offshore engineering) and models the headcount, cost, capacity and timing consequences against the current baseline.

     Who decides

     The CHRO, CFO and executive team choose which scenario to pursue; the agent builds and compares the options and flags where an assumption is weak.

     Metric

     planning-cycle time; quality of budget decisions (fewer mid-year re-plans)

     Data it needs

     headcount and cost baseline, comp ranges by level and location, attrition and hiring-velocity history, business plan drivers

   • Restructure Impact Analysis Agent Human decides restructure planning time; compliance defects; adverse-impact ratio of proposed pool

     What it does

     For a proposed restructure or reduction in force, calculates the roles in scope, cost and severance estimates, notice periods and consultation timelines by jurisdiction (WARN, collective-consultation thresholds), and drafts the workplan and adverse-impact analysis for the pool.

     Who decides

     Leadership, HR and legal decide the pool, the selection criteria and every individual outcome; the agent quantifies, checks timelines and flags adverse impact, and never selects or ranks anyone for redundancy.

     Metric

     restructure planning time; compliance defects; adverse-impact ratio of proposed pool

     Data it needs

     org and role data, comp and tenure, jurisdiction rules and thresholds, demographic data for adverse-impact analysis, aggregated

   In workforce planning, AI is strongest at forecasting, reconciliation and scenario maths on clean HRIS and finance data. Org-design judgment stays human, and most position data needs hygiene first.

   All 9 agents in this area

2. Talent Acquisition

   11 agents

   Talent Acquisition attracts, assesses and hires the people the workforce plan calls for, from outside and from within.

   7 proven · 2 strong case · 2 human decides

   JobsAttract candidates and present the employer brand · Build the pipeline before the req opens · Fill an open role · +5 more

   • Application Screening Support Agent Human decides recruiter screening hours; time-to-shortlist; adverse-impact ratio at screen

     What it does

     Reads each application against the published must-have criteria, summarizes the evidence for and against each criterion, and orders the pool for recruiter review, with reasons a recruiter can check and overturn.

     Who decides

     The recruiter or hiring manager decides who advances and who is declined, and sees every application, not only the top of the queue; the agent summarizes and orders for review and never auto-rejects. Candidates and, in the EU, workers' representatives are told the tool is in use; in NYC it is an automated employment decision tool requiring an annual bias audit and candidate notice, and in the EU it is high-risk under the AI Act.

     Metric

     recruiter screening hours; time-to-shortlist; adverse-impact ratio at screen

     Data it needs

     applications and resumes, published must-have criteria, bias-audit results by demographic group, candidate notice and consent records

   • Interview Scheduling Agent Proven time-to-schedule; interview-stage cycle time; recruiting-coordinator hours

     What it does

     Coordinates panel availability, candidate preferences, time zones, rooms and video links, sends invites and prep material, and handles reschedules and no-shows without a recruiter in the loop.

     Who decides

     The recruiter sets who interviews and in what order; the agent proposes and books slots and any human can override.

     Metric

     time-to-schedule; interview-stage cycle time; recruiting-coordinator hours

     Data it needs

     interviewer calendars, candidate availability, interview plan and panel, room and video-link systems

   • Structured Interview Support Agent Strong case interview consistency across panels; interviewer prep time; quality-of-hire proxies

     What it does

     Gives each interviewer a role-specific question guide tied to the competencies they own, avoids overlap across the panel, provides the scoring rubric, and after the interview turns their notes into a structured scorecard draft for them to correct and rate.

     Who decides

     Each interviewer records their own rating and recommendation; the agent prepares and drafts and does not score the candidate.

     Metric

     interview consistency across panels; interviewer prep time; quality-of-hire proxies

     Data it needs

     competency framework, success profile, interviewer notes or consented recordings, scorecard template

   In talent acquisition, AI is mature for job-ad drafting, scheduling, sourcing outreach and candidate communications. Anything that ranks candidates is regulated (NYC LL144, EU AI Act) and the hiring decision is never the machine's.

   All 11 agents in this area

3. Onboarding

   10 agents

   Onboarding takes a person from signed offer to productive, connected employee, through day one, the first 90 days and probation.

   4 proven · 5 strong case · 1 human decides

   JobsGet pre-boarding paperwork and compliance documents completed before day one · Have accounts, equipment, workspace and tool access ready on day one · Orient the new hire to the company, team and role · +5 more

   • New-Hire Question Answering Agent Proven ticket deflection rate and time-to-productivity

     What it does

     Answers the flood of first-weeks questions (how do I get expenses paid, where is the holiday policy, who approves my access request) from your handbook, benefits guides, IT documentation and team wikis, with citations, and hands off to a person when it is unsure.

     Who decides

     HR owns and approves the source documents; the AI answers only from them and escalates anything personal, sensitive or ambiguous to a named human.

     Metric

     ticket deflection rate and time-to-productivity

     Data it needs

     employee handbook and policy library, benefits and IT documentation, team-level wikis or FAQs, chat or email channel integration

   • Onboarding Process Health Agent Strong case day-one readiness rate, time-to-productivity and 90-day attrition

     What it does

     Watches the onboarding process, not the person: where paperwork stalls, which provisioning steps run late for which locations, where manager check-ins are not being scheduled, and which teams' new hires raise the most help requests, all aggregated across cohorts.

     Who decides

     The onboarding owner and the relevant IT, facilities or team leads decide process changes. The AI surfaces patterns at group level; it is not designed to monitor or evaluate any individual new hire or manager.

     Metric

     day-one readiness rate, time-to-productivity and 90-day attrition

     Data it needs

     step-level completion timestamps from onboarding tools, provisioning ticket data, check-in scheduling data, help-desk volumes tagged to new hires, aggregation and minimum-group-size rules

   • Probation Review Preparation Agent Human decides on-time probation reviews and 90-day attrition

     What it does

     Tracks probation milestones and jurisdiction-specific notice deadlines, reminds the manager and HR well ahead of each one, and assembles the manager's own notes, the agreed plan and check-in records into a structured template for the review conversation.

     Who decides

     The manager and HR decide whether to confirm, extend or end probation. The AI never scores the new hire, never recommends an outcome, and never drafts anything until the human decision is recorded.

     Metric

     on-time probation reviews and 90-day attrition

     Data it needs

     probation start and end dates by contract and jurisdiction, the 30-60-90 plan and check-in records the manager chose to keep, letter templates by country

   In onboarding, AI is strongest at coordination: chasing paperwork, tracking day-one provisioning, answering new-hire questions from your own documents. Any check-in signal stays at process level, never individual monitoring.

   All 10 agents in this area

4. HR Operations & Shared Services

   10 agents

   HR Operations & Shared Services answers employees' and managers' HR questions and runs the everyday transactions, records, documents and leave administration.

   5 proven · 4 strong case · 1 human decides

   JobsAnswer policy and process questions from your own documents · Take in HR cases and route them to the right person · Execute HRIS transactions and data changes correctly · +6 more

   • HR Policy Question Answering Agent Proven ticket deflection rate and first-contact resolution

     What it does

     Answers employee and manager questions (how much notice do I give, what counts as a dependant, how do I change my bank details) from your handbook, policies and process guides for the right country and entity, with citations, and hands off to a person when the question is personal or unclear.

     Who decides

     HR owns and approves the source documents and the hand-off rules. The AI answers only from those documents and escalates rather than guessing.

     Metric

     ticket deflection rate and first-contact resolution

     Data it needs

     policy and handbook library by country and entity, process guides and FAQs, employee attributes needed to pick the right policy version (location, entity, worker type), chat or portal integration

   • HRIS Transaction Preparation Agent Strong case transaction error rate and payroll rework

     What it does

     Turns a request in plain language (move this person to the new cost center from the 1st, record the title change agreed last week, update this address) into a pre-filled HRIS transaction, checks it against your business rules and effective-date logic, and flags anything that will break payroll or reporting.

     Who decides

     The HR operations specialist or the approving manager submits the transaction. The AI never commits changes to pay, grade or employment status; it prepares and validates.

     Metric

     transaction error rate and payroll rework

     Data it needs

     HRIS with write access behind an approval step, business rules, org structure and cost-center master data, effective-date and payroll cut-off calendar, request source (ticket, form or chat)

   • Policy Exception Request Preparation Agent Human decides consistency of exception handling and time-to-resolution

     What it does

     When someone asks for something outside policy (extra leave carry-over, a remote-work arrangement the policy does not cover, an unusual working-hours change), it gathers the relevant policy text, past exceptions of the same type at aggregate level, and the options open to HR, so the human has the full picture in one place.

     Who decides

     The HRBP or Head of People, with the manager, decides on the exception. The AI never recommends granting or refusing it and only drafts the reply after the decision is logged.

     Metric

     consistency of exception handling and time-to-resolution

     Data it needs

     policy library, exception log with categories and outcomes, employee context needed for the request (location, entity, role type)

   In HR operations, AI is the most mature of any HR area: document-grounded policy Q&A, ticket routing and letter generation are high-volume, well-understood work with clear payback.

   All 10 agents in this area

5. Compensation, Benefits & Payroll

   12 agents

   Compensation, Benefits & Payroll pays people accurately, administers benefits, and designs the pay structures and cycles that keep pay competitive, fair and explainable.

   5 proven · 5 strong case · 1 emerging · 1 human decides

   JobsRun payroll accurately and on time every cycle · Answer pay and benefits questions without a ticket queue · Get people enrolled in the right benefits · +5 more

   • Payroll Query Agent Proven ticket deflection rate, payroll query resolution time, HR and payroll hours reclaimed

     What it does

     Answers an employee's questions about their own payslip in plain language: why net pay changed, what a deduction line is, how overtime or a bonus was taxed, when a correction will land. Escalates anything it cannot explain from the payslip and policy to the payroll team with the context already gathered.

     Who decides

     The payroll team decides whether anything is wrong and whether to issue a correction; the agent explains what the payslip shows and routes suspected errors, it never adjusts pay.

     Metric

     ticket deflection rate, payroll query resolution time, HR and payroll hours reclaimed

     Data it needs

     employee's own payslip data, payroll calendar, pay and deduction policy documents, tax and statutory rules for the jurisdiction, identity verification via SSO

   • Compensation Band Drafting Agent Strong case time to refresh ranges, compa-ratio to target percentile, offer decline rate on compensation

     What it does

     Matches your jobs to market survey benchmarks, synthesizes survey data across sources and geographies, and drafts pay ranges by job family, level, and location for the total rewards team to review. Flags jobs whose current range has drifted from market and documents the match rationale for each.

     Who decides

     The head of total rewards decides the pay philosophy, target percentile, and final ranges, and the CHRO and CFO sign off on the structure; the agent drafts and shows its working.

     Metric

     time to refresh ranges, compa-ratio to target percentile, offer decline rate on compensation

     Data it needs

     licensed market survey data, job architecture and leveling framework, job descriptions, current ranges and incumbent pay, geographic differential policy

   • Individual Pay Options Agent Human decides pay-decision turnaround, off-cycle equity exceptions, regretted attrition citing compensation

     What it does

     For one person at a time (a merit increase, an off-cycle adjustment, a counter-offer, a new-hire salary) assembles the relevant facts: position in range, market reference, internal peers at level, tenure, and the rating the manager has given, then drafts an option set with the pay-equity and budget consequences of each.

     Who decides

     The manager proposes, total rewards reviews, and the approving leader decides the number. Nothing the agent produces is a recommendation to be accepted as-is, and it is never wired to write pay into the HRIS.

     Metric

     pay-decision turnaround, off-cycle equity exceptions, regretted attrition citing compensation

     Data it needs

     employee pay, level, position-in-range, tenure, peer pay at the same level and location, market reference from the pay bands, manager's rating and rationale, budget remaining

   In compensation and payroll, AI is strongest on high-volume, document-grounded work: payslip and benefits questions, pre-run payroll checks, enrollment guidance, market-data synthesis. Individual pay calls stay human.

   All 12 agents in this area

6. Performance & Talent Management

   10 agents

   Performance & Talent Management sets how goals, feedback, reviews, calibration, succession and promotion happen across the organization.

   2 proven · 6 strong case · 2 human decides

   JobsSet goals and keep them current · Make feedback continuous instead of annual · Run a fair, low-burden review cycle · +4 more

   • Goal Setting Support Agent Proven goal completion and update rates, share of employees with current goals, manager time on goal-setting

     What it does

     Helps employees and managers turn intentions into specific, measurable goals, shows how a draft goal lines up with team and company objectives, flags goals that have gone stale or have no progress signal, and prompts mid-quarter check-ins.

     Who decides

     The employee and their manager agree the goals; the agent drafts and flags.

     Metric

     goal completion and update rates, share of employees with current goals, manager time on goal-setting

     Data it needs

     company and team objectives, existing goals and progress updates, role profile and level expectations

   • Continuous Feedback Prompt Agent Strong case feedback frequency and coverage, feedback specificity, manager capability scores in engagement surveys

     What it does

     Prompts managers and peers to give feedback at natural moments (a project closes, a milestone lands, a 1:1 is coming up) and helps rewrite vague feedback into specific, behavior-based feedback before it is sent. Before a 1:1 it surfaces open goals, recent feedback, and prior action items as talking points.

     Who decides

     The feedback giver decides what to say and whether to send it; the manager decides what to raise in the 1:1. The agent prompts and helps phrase, it does not author feedback about someone on its own.

     Metric

     feedback frequency and coverage, feedback specificity, manager capability scores in engagement surveys

     Data it needs

     project and milestone signals from work tools, 1:1 and calendar data, goals and prior feedback, feedback quality guidelines

   • Promotion Case Drafting Agent Human decides promotion case preparation time, promotion rate consistency across cohorts, promotion appeals

     What it does

     When a manager nominates someone, assembles the evidence against the leveling criteria for the next level (scope, impact, behaviors), drafts the case narrative for the manager to edit, and marks which criteria have thin evidence. For the promotion committee, produces an aggregate view of nomination and proposed-approval rates by org and cohort.

     Who decides

     The manager decides whether to nominate; the promotion committee decides who is promoted; total rewards sets the pay outcome. The agent drafts and organizes evidence, it produces no readiness score or ranking of nominees.

     Metric

     promotion case preparation time, promotion rate consistency across cohorts, promotion appeals

     Data it needs

     leveling framework and promotion criteria, review outcomes, feedback and goal history, current level and time in level, manager's nomination rationale, aggregate nomination data for the consistency report

   In performance management, AI is useful on the burden side: turning notes into drafts, prompting feedback, assembling calibration packs, checking reviews for consistency. Ratings, promotions and underperformance calls stay with managers.

   All 10 agents in this area

7. Learning & Development

   10 agents

   Learning & Development finds the skills the organization is missing, gets the right learning to the right people, and shows the investment changed something.

   6 proven · 2 strong case · 1 emerging · 1 human decides

   JobsFind out where the skills gaps are · Get the right learning to the right person · Build and maintain the content library · +6 more

   • Skills Gap Analysis Agent Emerging skills coverage against strategic roles; time to complete a workforce skills review

     What it does

     Compares the skills a role or team needs against what people have declared, demonstrated in projects, or completed in learning, and shows where the gaps cluster. Runs continuously as roles and strategy change, so L&D is not rebuilding the picture from scratch each planning cycle.

     Who decides

     The employee confirms or disputes their own inferred profile; the L&D lead and business leaders decide which gaps to invest in. The AI infers, aggregates, and ranks gaps for review; it does not decide anyone's readiness for a role or assign anyone to work. Profiles are opt-in, visible to the employee and contestable.

     Metric

     skills coverage against strategic roles; time to complete a workforce skills review

     Data it needs

     role skill profiles or a skills taxonomy, learning history, self-declared skills, project or role history, optional: manager-validated skills

   • Compliance Training Tracking Agent Proven on-time completion rate; audit findings; HR hours reclaimed from chasing

     What it does

     Works out who owes which mandatory training based on role, location, and start date, sends timed reminders in the channel people read, escalates overdue items to the manager, and keeps an audit-ready record. Handles the recurring annual and role-change triggers so nobody has to run spreadsheets.

     Who decides

     The compliance or L&D owner sets the rules and decides what happens with persistent non-completion. The AI assigns by rule, nudges, and reports; any consequence for an individual is a manager or HR decision. Escalation rules are agreed with employee representatives where they exist.

     Metric

     on-time completion rate; audit findings; HR hours reclaimed from chasing

     Data it needs

     LMS completion records, HRIS role and location data, regulatory requirement matrix, messaging channel access

   • Program Nomination Support Agent Human decides panel time per selection round; consistency of decisions against stated criteria

     What it does

     Helps the panel that selects people for leadership programs, tuition sponsorship, or high-investment cohorts by assembling each nominee's stated goals, learning history, and manager nomination note against the published criteria, and by showing where the shortlist is skewed by function, level, or location.

     Who decides

     The selection panel decides who is offered a place; the AI organizes the evidence against the criteria and highlights gaps and skews. It does not score, rank, or recommend individuals for selection.

     Metric

     panel time per selection round; consistency of decisions against stated criteria

     Data it needs

     published program criteria, nomination forms, learning and skills history, aggregate demographic data for skew checks

   In L&D, AI is strongest on content: drafting, curating, tutoring, tracking, and recommending what to learn next. Inferring skills from people's work is much less settled and needs opt-in.

   All 10 agents in this area

8. Manager Enablement & Coaching

   10 agents

   Manager Enablement & Coaching makes sure every person who leads others can handle the situations in front of them, in the moment.

   5 proven · 4 strong case · 1 emerging

   JobsHelp a manager handle the situation in front of them · Prepare for and run good 1:1s · Rehearse the hard conversation before it happens · +7 more

   • Manager Coaching Agent Proven We built and ran this one manager capability; manager confidence in handling people situations; time from problem to action

     What it does

     Sits in Slack or Teams and coaches a manager through whatever they are facing right now: a report who has gone quiet, a peer conflict, a promotion request they cannot grant, a team that missed a deadline. Asks the questions a good coach would ask, offers frameworks and options, and helps the manager settle on an approach and the words to use.

     Who decides

     The manager decides what to say and do; the agent asks, suggests approaches, and drafts. Coaching conversations are private to the manager and are never shared with their own manager or HR. Only aggregate capability trends leave the system. Conversations are not retained as HR records.

     Metric

     manager capability; manager confidence in handling people situations; time from problem to action

     Data it needs

     a coaching and skills framework, company values and policies for grounding, chat platform integration, nothing from HRIS or performance records

   • Difficult Conversation Rehearsal Agent Strong case manager confidence going into hard conversations; time managers spend avoiding them; escalations to HR that could have been handled directly

     What it does

     Plays the other person in a conversation the manager is dreading, such as underperformance, a compensation disappointment, a redundancy notice, or a clash between two reports, at a chosen level of difficulty. Then gives feedback on clarity, empathy, and whether the manager said the hard thing.

     Who decides

     The manager decides whether and how to have the real conversation; HR and legal still own any formal process behind it. The AI simulates and gives feedback on the manager's approach; it never sees or produces anything about the real employee's record.

     Metric

     manager confidence going into hard conversations; time managers spend avoiding them; escalations to HR that could have been handled directly

     Data it needs

     scenario library and difficulty settings, coaching feedback rubric, company policy on the relevant process, nothing from HRIS or performance records

   • Manager Policy Q&A Agent Proven hRBP ticket deflection; manager time-to-answer; policy consistency

     What it does

     Answers the manager questions that otherwise queue in an HRBP's inbox: how do I approve extended leave, what is the process for a role change, can I give a spot bonus, what do I do if someone raises a grievance. Cites the policy, gives the steps, and hands off to the HRBP when the situation needs judgment.

     Who decides

     The manager acts within policy; the HRBP handles anything that requires discretion or involves a specific person's circumstances. The AI answers from documents and routes; it does not grant exceptions or interpret gray areas.

     Metric

     hRBP ticket deflection; manager time-to-answer; policy consistency

     Data it needs

     HR policy library, process documentation, HRBP routing rules, chat platform integration

   In manager coaching, in-the-flow chat coaching is the most proven build: a manager brings a real situation, gets a way through it, and acts the same day. Conversations stay private to the manager.

   All 10 agents in this area

9. Employee Engagement & Listening

   10 agents

   Employee Engagement & Listening hears what employees think and feel, turns it into action leaders take, and keeps people informed and recognized.

   3 proven · 6 strong case · 1 emerging

   JobsDesign and run the listening program (pulse, annual, lifecycle surveys) · Make sense of what employees say, at aggregate level · Turn survey insight into action plans leaders follow through on · +4 more

   • Open-Text Theme Synthesis Agent Proven time from survey close to results shared; share of leaders who receive a usable narrative

     What it does

     Reads thousands of free-text survey comments and groups them into themes with representative, de-identified quotes, always at a level where no individual or small team can be recognized, and shows how themes differ across large segments.

     Who decides

     The engagement lead and executive sponsor decide which themes become priorities and what is shared with whom; the agent summarizes and never surfaces comments below the group-size threshold or attributes a comment to a person.

     Metric

     time from survey close to results shared; share of leaders who receive a usable narrative

     Data it needs

     survey open-text responses, segment metadata at aggregate level, minimum group size rules, de-identification rules

   • Action Plan Drafting Agent Strong case share of teams with an action plan and visible progress; engagement index change year on year

     What it does

     For each leader whose team is above the reporting threshold, drafts two or three concrete commitments from their results and the themes raised, suggests how to run the results conversation, and follows up on progress at agreed intervals.

     Who decides

     The team leader chooses which commitments to make and owns them; the agent drafts options and tracks whether the leader has updated progress, escalating to the HRBP only when a plan is untouched. Escalation is a nudge agreed with the leader up front, not a record.

     Metric

     share of teams with an action plan and visible progress; engagement index change year on year

     Data it needs

     team-level survey results above threshold, theme summaries, prior action plans, calendar or task tool access

   • Internal Communications Drafting Agent Proven comms turnaround time; HR and comms hours reclaimed

     What it does

     Drafts announcements, town-hall scripts, leader talking points, FAQs and change communications from a brief, adapts tone and length for different audiences and languages, and checks drafts against approved terminology and prior messaging.

     Who decides

     The internal comms lead or executive sponsor approves and sends; the agent drafts and flags inconsistencies, it does not publish.

     Metric

     comms turnaround time; HR and comms hours reclaimed

     Data it needs

     approved messaging and terminology, prior communications archive, brief for the announcement, audience segments and languages

   In employee listening, AI is strongest at reading open-text comments at scale and drafting the follow-up plans and communications. It only works aggregate-only, above minimum group sizes.

   All 10 agents in this area

10. Employee Relations, Policy & Compliance

    10 agents

    Employee Relations, Policy & Compliance keeps policies current, handles concerns, grievances and investigations with due process, and works with works councils and unions.

    2 proven · 6 strong case · 2 human decides

    JobsAnswer employees' and managers' policy questions accurately · Keep policies current, consistent and versioned across jurisdictions · Take in ER concerns and get them to the right person fast · +5 more

    • Policy Drafting & Version Control Agent Strong case policy update cycle time; number of out-of-date or conflicting policies

      What it does

      Drafts new or updated policies from templates and a regulatory change brief, produces redlines against the current version, checks for conflicts with other policies and collective agreements, and keeps a version history with effective dates per jurisdiction.

      Who decides

      The Head of People and legal counsel approve the policy, and the works council is consulted where required; the agent drafts and flags conflicts, it does not publish a policy.

      Metric

      policy update cycle time; number of out-of-date or conflicting policies

      Data it needs

      current policy library, policy templates, collective agreements and works agreements, regulatory change briefs, document management access

    • ER Case Intake & Triage Agent Strong case time from concern raised to human contact; case backlog; consistency of routing

      What it does

      Takes a concern raised by an employee or manager through a structured, respectful intake, captures the facts as stated, classifies the category and urgency (for example harassment, conduct, wellbeing, pay dispute), and routes it to the right ER partner, legal or safeguarding contact within the required time; it does not comment on merits.

      Who decides

      The ER partner reviews the classification, decides how the case will be handled and who investigates; the agent captures and routes, and never advises the reporter on the strength of their concern. The reporter can reach a person at any point.

      Metric

      time from concern raised to human contact; case backlog; consistency of routing

      Data it needs

      case management system, routing rules by category, location and entity, conflict-of-interest data (reporting lines), statutory response timelines

    • Disciplinary Process Support Agent Human decides procedural compliance on audit; consistency of outcomes across similar cases; disputes lost on process grounds

      What it does

      Checks that each step of the disciplinary procedure has been followed before the next one starts, drafts the invitation letter once HR confirms there is a case to answer and the outcome letter once the hearing manager has recorded their decision, and pulls together the anonymized precedent range so the decision-maker can check for consistency.

      Who decides

      The disciplinary hearing manager decides whether there is a case to answer and what sanction, if any, applies, with HR advising; the agent checks process, drafts the letter for the human's stated decision, and never recommends a sanction.

      Metric

      procedural compliance on audit; consistency of outcomes across similar cases; disputes lost on process grounds

      Data it needs

      disciplinary procedure and templates, case management system, anonymized prior case outcomes, employment contract and collective agreement terms

    In employee relations, AI is useful for document-heavy, deadline-heavy work: policy Q&A, drafting and versioning policies, watching regulation, keeping investigation files complete. Every conclusion and sanction stays with a person.

    All 10 agents in this area

11. People Analytics & Reporting

    10 agents

    People Analytics & Reporting turns HRIS, ATS, LMS, payroll and survey data into trustworthy numbers and plain-language answers leaders can use.

    2 proven · 8 strong case

    JobsAnswer a people question fast · Report headcount, cost and movement on a cadence · Prepare board and executive people reporting · +6 more

    • People Data Query Agent Strong case time-to-insight; analyst hours reclaimed from ad-hoc requests

      What it does

      Lets an HRBP or leader ask a question in plain language (how many engineers did we lose in EMEA last quarter, by tenure band?) and translates it into a query against the governed people data model, returning the answer with the definitions and filters it used.

      Who decides

      The requester decides what to do with the answer; the People Analytics team owns the data model and definitions the agent is allowed to use. The agent answers within role-based access limits and suppresses small cells; it never releases row-level data the requester could not already see.

      Metric

      time-to-insight; analyst hours reclaimed from ad-hoc requests

      Data it needs

      a governed people data model or warehouse view, metric dictionary, role-based access rules, small-cell suppression thresholds

    • Headcount & Cost Reporting Agent Proven report cycle time; HR hours reclaimed; reconciliation error rate

      What it does

      Assembles the weekly or monthly people pack: headcount and FTE by org, hires, exits, transfers, open reqs, employment cost against plan, reconciled to Finance's numbers, and drafts the variance commentary.

      Who decides

      The People Analytics lead or HR ops lead reviews and signs off the pack before it goes to leadership; the agent assembles, reconciles and drafts, and flags where HR and Finance numbers disagree rather than picking one.

      Metric

      report cycle time; HR hours reclaimed; reconciliation error rate

      Data it needs

      HRIS position and worker data, payroll or finance cost data, headcount plan and budget, ATS open requisitions, org hierarchy

    • Retention Investment Review Agent Strong case regretted attrition rate in targeted segments; retention program return

      What it does

      Turns cohort attrition drivers into a segment-level watchlist (for example, engineers at 18 to 30 months tenure below pay-band midpoint) with the estimated cost of losing versus retaining that segment, so HR can decide where stay-interview programs, pay-band corrections or career-path work should go.

      Who decides

      HRBPs and the CHRO decide which segments to invest in; any individual action (a stay conversation, a retention adjustment, a counteroffer) is decided by the manager and HR with the person, never triggered by a score. The agent ranks segments above a minimum group size and drafts options; it does not name or score individuals.

      Metric

      regretted attrition rate in targeted segments; retention program return

      Data it needs

      cohort driver analysis output, compensation and pay-band data, replacement-cost assumptions, open req and time-to-fill data

    In people analytics, AI is strongest on the plumbing: cross-system data-quality checks, cadence reporting, plain-language answers over a curated data model. Everything stays at cohort grain, never individual scores.

    All 10 agents in this area

12. Offboarding & Alumni

    10 agents

    Offboarding & Alumni runs every departure cleanly and lawfully, keeps leavers' knowledge and goodwill, and turns alumni into insight, referrals and rehires.

    4 proven · 5 strong case · 1 human decides

    JobsRun a clean, compliant exit from notice to last day · Learn from leavers · Keep the knowledge when the person goes · +6 more

    • Exit Process Coordination Agent Proven HR hours per exit; on-time exit-checklist completion; leaver experience score

      What it does

      When a resignation or separation is entered, builds the exit plan for that person's jurisdiction, role and system footprint, opens tasks for the manager, IT, Finance, Payroll and Facilities, chases what is late, and answers the leaver's process questions.

      Who decides

      The HR ops owner decides exceptions (early release, garden leave, a disputed last day); the manager and HR decide anything about the person. The agent orchestrates tasks and reminders and flags stalls; it does not change employment status.

      Metric

      HR hours per exit; on-time exit-checklist completion; leaver experience score

      Data it needs

      HRIS separation record, exit checklist templates by country and role, task or ticket system access, calendar, system and asset inventory

    • Exit Interview Synthesis Agent Strong case regretted attrition (aggregate); repeat-theme resolution rate; exit-survey response rate

      What it does

      Collects exit-survey and exit-interview notes (or runs a structured, opt-in exit conversation itself), then synthesizes themes by function, level, tenure and manager span at aggregate level, tracking how reasons for leaving shift quarter to quarter.

      Who decides

      The CHRO and HRBPs decide what to act on; the agent synthesizes and reports at aggregate. Comments about a named manager or a possible misconduct disclosure are routed to HR or ER for a human to read, not summarized into a score. Leavers are told this before they start.

      Metric

      regretted attrition (aggregate); repeat-theme resolution rate; exit-survey response rate

      Data it needs

      exit survey responses, exit interview notes or transcripts (with consent), HRIS attributes for segmentation, minimum group-size rules

    • Alumni Rehire Matching Agent Human decides boomerang hire rate; time-to-fill and time-to-productivity on rehires

      What it does

      When a requisition opens, matches it against opted-in alumni by prior role, skills, tenure and stated interests, and produces a shortlist for the recruiter to reach out to, showing the rehire-eligibility status HR recorded at exit.

      Who decides

      Rehiring is a hiring decision: the recruiter decides who to contact, and the hiring manager and HR decide any offer. Rehire eligibility is set by HR at exit, not by the agent; the agent only surfaces the recorded status and ranks for review. Because it ranks people for a role, it carries the same candidate-notice and bias-audit obligations as screening (NYC LL144, EU AI Act high-risk).

      Metric

      boomerang hire rate; time-to-fill and time-to-productivity on rehires

      Data it needs

      alumni profiles with consent, open requisitions and required skills, rehire-eligibility flag from HRIS, ATS

    In offboarding, AI is already reliable for coordination and documents: checklists across HR, IT and Finance, deprovisioning, letters, verifications, and aggregate exit-interview themes.

    All 10 agents in this area