All 8 run on whichever AI agent you already use -

Microsoft Copilot - connected to the tools listed with each use case.

| # | Use case | Key integrations | Est. time saved About the estimate | | - | ------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | 1 | Research candidates |

Talent sourcing

ATS | **2.5** hr/wk | | 2 | Answer policy questions |

Internal knowledge

Meetings

ATS | **2** hr/wk | | 4 | Draft job descriptions |

ATS

HRIS

HRIS

BI | **1.5** hr/wk | | 7 | Analyze employee surveys |

Surveys

HRIS

Docs | **1** hr/wk |

Research candidates

Best for assembling role-relevant evidence before recruiter review

2.5 hr/wkest. time saved How this estimate is calculated

How it's done today

Recruiters search profiles and work samples, compare them with role criteria, record uneven notes, and lose source context while moving prospects into the ATS.

How AI helps

Your agent gathers permitted public professional evidence, maps it to the approved criteria, links every observation to its source, and prepares a research brief without making the hiring decision.

Candidate enters list approved role criteria available

Your AI agent

Finds role-relevant professional evidence, distinguishes observed facts from inference, and records gaps rather than guessing.

Candidate brief

Source links

Recruiter questions

How to set it up

Step 1Connect your tools

Required

Talent sourcing

Reads public professional profiles and work evidence allowed by policy.

ATS

Reads the role, approved criteria, candidate status, and duplicate records.

Writes a source-linked research note for recruiter review.

Optional

Docs

Reads the scorecard, sourcing policy, and example evidence standards.

Provide the role scorecard and candidate URLs manually. Do not ask the agent to search or infer protected or personal information.

Step 2Build the workflow

Paste this into your agent or automation tool. Have the approved scorecard, sourcing policy, allowed-source list, one strong research brief, and three real profiles representing clear evidence, adjacent experience, and too little evidence ready.

```text Setup prompt theme={null} Help me set up a candidate-research workflow. It should prepare a source-linked brief against an approved role scorecard. It must never rank, reject, or advance a candidate. 1. Ask me which sourcing and ATS tools I use and which public sources are allowed. 2. Ask me for the approved scorecard and the observable evidence for each criterion. 3. Ask me which attributes and sources are prohibited and what must be omitted. 4. Ask me for one strong research brief to use as the output example. 5. Build the workflow so every criterion lists observed evidence, source URL, source date, confidence, and what remains unknown. 6. Keep fact separate from inference, ignore protected or personal information, and produce recruiter questions rather than a hiring recommendation. 7. Test it on my three real profiles and show me the brief and evidence gaps. ```

What good looks like

Across the three profiles, every statement should trace to permitted professional evidence, the same scorecard should be applied consistently, and missing evidence should stay missing. Remove any unsupported inference or prohibited detail and rerun the same profiles before using the workflow.

Step 3Turn it on

Choose your trigger

Run it when a recruiter adds a profile to an approved sourcing project with a selected role and scorecard. Keep internal candidates, restricted jurisdictions, and unapproved sources outside this workflow.

What runs without you

The agent can gather allowed evidence and prepare an ATS note automatically, but a recruiter reviews the brief and owns every outreach or hiring action. Sample the generated notes weekly for unsupported inference, source drift, and prohibited information, and disable any source that stops meeting the policy.

Pairs well with summarize interview evidence - both build the evidence pack the hiring team decides from.

Answer policy questions

Best first workflow - give employees a cited answer at the point of need

2 hr/wkest. time saved How this estimate is calculated

How it's done today

Employees search several intranet pages or message HR, who checks location, worker type, and policy version before rewriting an answer and linking the source.

How AI helps

Your agent identifies the applicable policy scope, retrieves the current passage, answers in plain language with a citation, and routes personal, ambiguous, or exception requests to HR.

Employee asks policy question in chat or portal

Your AI agent

Matches the question to current approved policy, applies location and worker type, and cites the exact passage or creates a human handoff.

Cited answer

Policy source

HR handoff

How to set it up

Step 1Connect your tools

Required

Internal knowledge

Reads approved policies, effective dates, locations, worker types, owners, and archived versions.

Recommended

Chat

Reads the employee question and chosen scope fields.

Writes the cited answer or private handoff link.

Optional

HRIS

Reads only the employment type and location needed to choose the applicable policy.

Employees can ask the agent directly against a curated policy collection and select their location and worker type manually.

Step 2Build the workflow

Paste this into your agent or automation tool. It interviews you, builds the workflow, and tests it before anything goes live - have your current policy collection, one answer HR considers excellent, and three real questions covering a simple answer, a scope difference, and a necessary handoff ready.

```text Setup prompt theme={null} Help me set up an HR policy-answer workflow. It should answer routine employee questions from current policy and create a useful private handoff when HR needs to respond. 1. Ask me where approved policies live and which chat or portal employees use. 2. Ask me how policy versions, locations, worker types, and owners are recorded. 3. Ask me which questions must always go to HR and what the handoff should contain. 4. Ask me for one good cited answer to use as the format example. 5. Build the workflow so it identifies scope, retrieves the controlling passage, and answers with the policy title, effective date, quotation, and link. 6. Never infer eligibility or exceptions. If scope is missing, sources conflict, or the case is personal, prepare a private handoff with the sources checked. 7. Test it on my three real questions and show me the answer or handoff for each. ```

What good looks like

Across the three tests, the simple question should receive the right current citation, the scoped question should use the correct location and worker type, and the personal case should become a useful private handoff. Correct any wrong scope, stale source, or vague escalation and rerun the same three questions.

Step 3Turn it on

Choose your trigger

Run it when an employee submits a question through the approved HR portal or private chat entry point. Keep public channels and messages containing personal details outside the answering flow and send them directly to a private HR handoff.

What runs without you

The agent can answer routine questions whose scope and source are unambiguous. HR continues to review personal cases, conflicts, exceptions, and anything without a current controlling passage. Once a month, check unanswered questions and expiring policies so the workflow does not quietly rely on stale guidance.

Pairs well with create onboarding plans - new hires generate most of the questions the answer flow fields.

Summarize interview evidence

Best for turning several interviews into a comparable evidence pack

2 hr/wkest. time saved How this estimate is calculated

How it's done today

Interviewers leave notes in different formats, recruiters chase scorecards, and the hiring team rereads transcripts while the strongest personality can outweigh the stated criteria.

How AI helps

Your agent maps notes and transcripts to the approved scorecard, cites observed evidence and gaps, and prepares a comparable packet without recommending hire or no-hire.

Interview round closes notes and scorecards submitted

Your AI agent

Organizes interview evidence by criterion, preserves source and interviewer, surfaces conflicts, and leaves the decision to the hiring team.

Evidence packet

Scorecard gaps

Debrief brief

How to set it up

Step 1Connect your tools

Required

Meetings

Reads consented transcripts, notes, speakers, and timestamps.

ATS

Reads the scorecard, interview plan, submitted ratings, and candidate stage.

Writes the evidence packet and missing-scorecard flags.

Optional

Docs

Reads calibration examples and debrief template.

Upload interviewer notes and completed scorecards; a transcript is useful but not required. Use candidate and interviewer IDs consistently.

Step 2Build the workflow

Paste this into your agent or automation tool. Have the structured scorecard, consent rules, one useful debrief packet, and three completed interview rounds covering consistent evidence, interviewer disagreement, and a missing scorecard ready.

```text Setup prompt theme={null} Help me set up an interview-evidence synthesis workflow. It should prepare a comparable debrief packet after interviewers submit their independent scorecards. It must never recommend the hiring decision. 1. Ask me which meeting and ATS tools hold transcripts, notes, and scorecards. 2. Ask me for the scorecard, evidence standard, consent rules, and debrief format. 3. Ask me when a round is complete and how missing scorecards should be handled. 4. Ask me for one debrief packet that shows the right level of evidence and detail. 5. Build the workflow so each criterion shows supporting and contradicting evidence with interviewer, source, timestamp, original rating, and missing assessment. 6. Preserve ratings, flag disagreements, omit protected information, and produce debrief questions rather than changing scores or choosing a candidate. 7. Test it on my three completed rounds and show me the packet for each. ```

What good looks like

The three packets should map evidence to the right criteria, preserve original ratings, expose disagreement, and leave unassessed criteria visibly incomplete. Correct any lost source, flattened disagreement, or inferred trait and rerun the same rounds.

Step 3Turn it on

Choose your trigger

Run it only after the interview round closes and all available independent scorecards have been submitted. Never expose other interviewers’ ratings before submission or include a recording that was not approved for this use.

What runs without you

The agent can assemble the evidence packet and notify the recruiting team when it is ready. Recruiters and interviewers still conduct the debrief and own the decision. Each month, sample packets for missing scorecards, evidence that lost its timestamp, and information that should not have entered the workflow.

Pairs well with research candidates - pre-interview evidence and interview evidence belong in one packet.

Draft job descriptions

Best for turning an approved hiring request into consistent candidate-facing copy

1.5 hr/wkest. time saved How this estimate is calculated

How it's done today

Recruiters combine a hiring request, leveling guide, competencies, location, compensation rules, and recycled descriptions, then remove inflated or inconsistent requirements.

How AI helps

Your agent maps the approved role inputs into the company template, distinguishes required from preferred qualifications, and flags missing approvals or inconsistent levels.

Requisition approved level and location set

Your AI agent

Builds a clear role description from approved inputs, checks consistency and inclusive language, and leaves approval gaps visible.

Job description

Requirements check

Posting draft

How to set it up

Step 1Connect your tools

Required

ATS

Reads the requisition, level, location, hiring team, and posting fields.

Writes a job-posting draft only.

Docs

Reads the template, leveling guide, competencies, approved benefits, and language rules.

Writes the reviewable description.

Optional

HRIS

Reads approved job family and compensation-band metadata.

Upload the requisition, template, and leveling guide. The agent can create the draft without ATS access; posting still remains manual.

Step 2Build the workflow

Paste this into your agent or automation tool. Have an approved requisition, your current job-description template, two descriptions recruiters consider strong, and three real openings from different role families ready for the setup interview and test.

```text Setup prompt theme={null} Help me set up a job-description drafting workflow. It should turn an approved hiring request into a candidate-facing ATS draft. It must never post the job or invent requirements. 1. Ask me which ATS and document tools I use and confirm where drafts should land. 2. Ask me for the approved template, leveling guide, location rules, and company copy. 3. Ask me for two strong descriptions and how required and preferred skills differ. 4. Ask me which requisition fields and approvals must exist before drafting starts. 5. Build the workflow so it uses the approved outcomes, responsibilities, level, location, compensation rules, and qualifications to create the draft. 6. If inputs conflict or are missing, return a specific gap list instead of filling them in. Save only an ATS or Docs draft for recruiter and manager review. 7. Test it on my three real openings and show me the draft and gaps for each. ```

What good looks like

Each test draft should match the approved level and actual work, separate true requirements from preferences, use the correct location and compensation language, and expose missing approvals. Fix any inflated qualification, inconsistent title, or copied contradiction and rerun the same three openings.

Step 3Turn it on

Choose your trigger

Run it when a requisition moves to Approved and has an owner, level, location, outcomes, and required approvals. Incomplete requisitions should stop at a gap list rather than creating generic copy.

What runs without you

The workflow can create the first ATS draft and notify the recruiter automatically. Recruiters and hiring managers still approve requirements, compensation language, and the final posting. Review templates and location rules each quarter, and sample drafts whenever a new role family is added.

Pairs well with research candidates - the same scorecard defines the posting and the research criteria.

Create onboarding plans

Best for making every new hire’s first weeks complete and role-specific

1.5 hr/wkest. time saved How this estimate is calculated

How it's done today

HR and managers copy a checklist, look up team, role, location, equipment, training, and meetings, then discover readiness gaps only days before the start date.

How AI helps

Your agent combines approved new-hire and role inputs with the onboarding template, builds a dated plan, checks readiness, and routes missing owners or prerequisites before day one.

Hire marked accepted start date and manager confirmed

Your AI agent

Selects the correct role, location, and team tasks, creates milestones, and flags equipment, access, owner, or scheduling gaps.

Onboarding plan

Readiness check

Manager brief

How to set it up

Step 1Connect your tools

Required

HRIS

Reads approved cohort, name, role, team, manager, location, employment type, and start date.

Tasks

Reads the onboarding template and owners.

Writes the new plan, tasks, dependencies, and milestones.

Recommended

Calendar

Writes draft orientation and check-in holds after manager review.

Optional

Internal knowledge

Reads approved team guides and training resources.

Use an approved intake form and create the plan from a template. Avoid copying sensitive HR fields that the onboarding task owners do not need.

Step 2Build the workflow

Paste this into your agent or automation tool. Have approved onboarding templates, owner and lead-time rules, one plan the team considers complete, and three recent hires covering a standard start, a remote start, and a late-notice start ready.

```text Setup prompt theme={null} Help me set up a new-hire onboarding workflow. It should turn an accepted hire record into a complete, reviewable onboarding plan without exposing employee information people do not need. 1. Ask me which HRIS, task, calendar, and document tools I use. 2. Ask me for the approved templates by role, location, and employment type. 3. Ask me who owns equipment, access, training, meetings, buddy setup, and goals. 4. Ask me for lead times, privacy rules, and one onboarding plan to copy structurally. 5. Build pre-start, day-one, week-one, 30-, 60-, and 90-day tasks with owners, due dates, dependencies, links, and only the employee fields each owner needs. 6. Flag missing owners and impossible dates; stage tasks and calendar holds for review rather than silently skipping them. 7. Test it on my three recent hires and show me the plan and readiness gaps. ```

What good looks like

Each test plan should select the right role and location template, give every task an owner and achievable date, and expose missing access, equipment, or training before the start date. Fix any wrong template, owner, or dependency and rerun the same three hires.

Step 3Turn it on

Choose your trigger

Run it when an accepted hire has a start date, manager, role, location, and employment type. Exclude rescinded and duplicate records, and do not create tasks until the hire is approved for onboarding.

What runs without you

The workflow can create standard operational tasks and reminders automatically. Managers still approve goals, role-specific meetings, and exceptions, while HR owns sensitive changes. Send owners a weekly readiness digest until the start date and flag any overdue dependency rather than repeatedly creating tasks.

Pairs well with answer policy questions - a good plan prevents half the policy questions, and the answer flow catches the rest.

Analyze workforce trends

Best for recurring workforce reports with consistent definitions

1.5 hr/wkest. time saved How this estimate is calculated

How it's done today

HR analysts reconcile headcount, hiring, movement, tenure, absence, and attrition across snapshots, definitions, and org changes before explaining the movement.

How AI helps

Your agent refreshes approved workforce measures, validates populations and snapshots, decomposes material changes, and drafts a privacy-safe report with evidence and owner questions.

Reporting period closes HR snapshot is complete

Your AI agent

Applies governed workforce definitions, reconciles snapshots and movements, and explains material changes without profiling individuals.

Workforce brief

Metric table

Owner questions

How to set it up

Step 1Connect your tools

Required

HRIS

Reads approved workforce snapshots, movements, org structure, and governed dimensions.

Reads metric definitions, targets, trends, and privacy-safe views.

Writes the internal report draft.

Optional

Docs

Reads the reporting template and prior decisions.

Writes the narrative and open questions.

Use a privacy-protected aggregate export and the workforce metric dictionary. Avoid person-level rows unless the approved analysis genuinely requires them.

Step 2Build the workflow

Paste this into your agent or automation tool. Have the metric dictionary, certified workforce snapshots, org-change mapping, one approved report, and three periods covering stable movement, a reorganization, and a material attrition change ready.

```text Setup prompt theme={null} Help me set up a workforce-trend reporting workflow. It should reconcile certified workforce movement and prepare a privacy-safe internal report. It must not speculate about individual employees. 1. Ask me which HRIS and BI tools hold the certified snapshots and report draft. 2. Ask me for metric definitions, snapshot dates, movement rules, and control totals. 3. Ask me for org-change mapping, privacy thresholds, targets, and materiality rules. 4. Ask me for one approved report to use as the structure and language example. 5. Build the workflow so opening headcount plus hires, transfers, and exits ties to closing headcount before it calculates trends or segment drivers. 6. Separate observed movement from possible explanations, suppress unsafe groups, and create owner questions wherever the data does not support a cause. 7. Test it on my three real periods and show me the reconciliation and report. ```

What good looks like

All three periods should reconcile, the reorganization should use the approved mapping, small groups should remain protected, and unsupported explanations should become owner questions. Correct any broken control, inconsistent definition, or unsafe segment and rerun the same periods.

Step 3Turn it on

Choose your trigger

Run it after the certified HR snapshot closes and the org mapping is approved. Keep provisional movement and groups below the privacy threshold out of the narrative and route them to a private exception view.

What runs without you

The agent can refresh the reconciled tables and prepare the internal narrative draft on each certified snapshot. HR reviews interpretation and distribution, especially after reorganizations. Keep the reconciliation result with every report and stop the workflow automatically whenever a control does not tie.

Pairs well with analyze employee surveys - the numbers say what changed and the surveys suggest why.

Analyze employee surveys

Best for turning open text and scores into themes leaders can act on

1 hr/wkest. time saved How this estimate is calculated

How it's done today

HR exports scores and comments, cleans segments, reads open text, codes themes, suppresses small groups, and writes a summary while trying not to expose individuals.

How AI helps

Your agent applies privacy thresholds, calculates approved comparisons, groups comments into evidence-backed themes, and drafts a report with representative de-identified examples and action questions.

Survey closes response and privacy thresholds met

Your AI agent

Validates the population, suppresses small groups, calculates approved changes, and synthesizes de-identified themes with counts.

Survey brief

Theme table

Action questions

How to set it up

Step 1Connect your tools

Required

Surveys

Reads responses, question text, scale definitions, dates, and permitted comment fields.

Recommended

HRIS

Reads approved segment fields through privacy-protected analysis views.

Optional

Writes aggregated, suppressed results and trend views.

Export an aggregated dataset with small groups already suppressed. Do not give the workflow identifiers it does not need.

Step 2Build the workflow

Paste this into your agent or automation tool. Have the survey export, metric and privacy rules, the theme taxonomy, one approved report, and three representative slices covering healthy volume, a suppressed small group, and polarized comments ready.

```text Setup prompt theme={null} Help me set up an employee-survey analysis workflow. It should produce a privacy-safe findings draft from a closed survey. It must not identify respondents or expose a group below the approved threshold. 1. Ask me which survey and HR tools I use and where the approved export lives. 2. Ask me for scale direction, comparison periods, segment rules, and control totals. 3. Ask me for minimum-group, comment-suppression, and identifying-text rules. 4. Ask me for the theme taxonomy and one approved report to use as the example. 5. Build the workflow so it validates counts, calculates distributions and changes, and groups de-identified comments into themes with counts and examples. 6. Preserve minority themes, suppress unsafe cuts, and separate observed results from possible explanations or recommended questions. 7. Test it on my three representative slices and show me the output for each. ```

What good looks like

The test outputs should reconcile to survey totals, suppress the small group, retain meaningful minority themes, and avoid identifying language. Correct any reversed scale, unsafe segment, or unsupported theme and rerun the same three slices.

Step 3Turn it on

Choose your trigger

Run it only after the survey closes and the export passes the agreed privacy checks. Remove identifying free text and blocked small-group cuts before the agent receives the analysis file.

What runs without you

The workflow can refresh approved aggregate tables and prepare a private report draft. HR reviews interpretation, examples, and distribution every cycle. Reconfirm privacy thresholds and segment definitions before each survey, and log any question or cut the workflow had to suppress.

Pairs well with analyze workforce trends - survey themes explain the movements the workforce report finds.

Prepare performance reviews

Best for organizing the review period’s evidence into a consistent draft

1 hr/wkest. time saved How this estimate is calculated

How it's done today

Managers reconstruct months of goals, feedback, outcomes, and development notes, then write at different levels of detail and recency while HR checks completeness and consistency.

How AI helps

Your agent assembles approved review-period evidence into the company template, maps examples to competencies and goals, and flags unsupported statements or missing input for the manager.

Review window opens goals and evidence cutoff reached

Your AI agent

Organizes documented outcomes and feedback by goal and competency, checks evidence coverage, and drafts language for manager review.

Review draft

Evidence map

Missing input

How to set it up

Step 1Connect your tools

Required

HRIS

Reads the review template, goals, competencies, review period, and approved feedback.

Writes a private draft for the manager.

Recommended

Docs

Reads documented work outcomes and development notes within the review period.

Optional

Tasks

Reads completed goals and project outcomes when approved as evidence.

Managers can upload the template and an approved evidence packet. Do not let the agent search private communications broadly for performance evidence.

Step 2Build the workflow

Paste this into your agent or automation tool. Have the current review template, evidence and calibration rules, one strong completed review, and three past evidence packs covering complete, sparse, and conflicting records ready.

```text Setup prompt theme={null} Help me set up a performance-review preparation workflow. It should organize approved review-period evidence and prepare a manager draft. It must never choose a rating or make an employment decision. 1. Ask me which HRIS and document tools hold goals, feedback, and review forms. 2. Ask me for the review period, competencies, evidence rules, and template. 3. Ask me which sources are allowed and which private messages must stay excluded. 4. Ask me for one strong review that shows specific evidence-based language. 5. Build the workflow so each statement maps to a goal or competency with its source and date, while results, manager judgment, and future goals stay distinct. 6. Flag missing employee input, unsupported ratings, recency gaps, and conflicts; leave the rating and final wording to the manager. 7. Test it on my three past evidence packs and show me each draft and gap list. ```

What good looks like

Across the three tests, every material statement should trace to review-period evidence, sparse records should remain visibly sparse, and conflicting evidence should not be flattened into certainty. Fix unsupported wording or misplaced evidence and rerun the same packs.

Step 3Turn it on

Choose your trigger

Run it after the review-period evidence cutoff and before the manager begins final drafting. Keep unapproved private messages and material outside the review period out of the source set.

What runs without you

The workflow can assemble evidence and create the first review draft automatically. The manager remains the author, sets the rating, resolves conflicts, and approves every review. At the start of each cycle, audit source permissions and confirm that old review-period evidence is not being carried forward.

Pairs well with analyze workforce trends - both depend on the same review-period evidence being complete.