--- title: "Staffing AI rules: data, approvals, and spend" description: "Set staffing AI rules for candidate records, hiring decisions, client messages, and function budgets, with approval owners and checked primary sources" canonical: "https://scalewithsearch.com/articles/staffing-ai-rules-data-approvals-spend" --- ## Site navigation - [Scale With Search](https://scalewithsearch.com/) - Services - Services - [AI systems](https://scalewithsearch.com/ai-systems) - [Site build](https://scalewithsearch.com/work#site-build) - [Whole website fix](https://scalewithsearch.com/work#whole-website-fix) - [Content library with SEO](https://scalewithsearch.com/work#content-library) - [Working session](https://scalewithsearch.com/audit) - Start here - [Send your brief](https://scalewithsearch.com/work#send-your-brief) - [Example scope](https://scalewithsearch.com/work#review-the-example-scope) - [Prices and terms](https://scalewithsearch.com/work#prices-and-terms) - Industries - Real estate - [Real estate agents and teams](https://scalewithsearch.com/for/real-estate) - [Platforms, ownership, and IDX](https://scalewithsearch.com/for/real-estate/real-estate-website-platforms-ownership-idx) - [Search, content, and AI](https://scalewithsearch.com/for/real-estate/real-estate-search-content-ai) - [Inquiries, campaigns, and handoffs](https://scalewithsearch.com/for/real-estate/real-estate-website-inquiries-campaigns-handoffs) - [Buyer, seller, and local pages](https://scalewithsearch.com/for/real-estate/real-estate-buyer-seller-local-resources) - [Rules, privacy, and trust](https://scalewithsearch.com/for/real-estate/real-estate-website-rules-privacy-trust) - Browse by business - [Trades and home services](https://scalewithsearch.com/for/#group-trades) - [Health, wellness, and animal care](https://scalewithsearch.com/for/#group-practices) - [Law, finance, and IT](https://scalewithsearch.com/for/#group-professional) - [Marketing and SEO agencies](https://scalewithsearch.com/for/#group-agencies) - [Shops and manufacturers](https://scalewithsearch.com/for/#group-manufacturing) - [Restaurants, shops, studios, and nonprofits](https://scalewithsearch.com/for/#group-local-services) - Chiropractic practices - [Chiropractic and wellness practices](https://scalewithsearch.com/for/chiropractic-and-wellness-practices) - [Chiropractic and physical therapy clinics](https://scalewithsearch.com/for/chiropractic-and-physical-therapy-clinics) - [All industries](https://scalewithsearch.com/for/) - Resources - Free tools - [AI readiness check](https://scalewithsearch.com/check) - [Free tools](https://scalewithsearch.com/tools/) - [Free Starter Kit](https://scalewithsearch.com/kit/business-memory-starter-kit.zip) - Guides - [AI-native library](https://scalewithsearch.com/ai-native) - [Industry intelligence](https://scalewithsearch.com/intelligence) - [Office job guides](https://scalewithsearch.com/guides/) - [Business memory](https://scalewithsearch.com/articles/business-memory-for-ai-agents-guide) - [Agent governance](https://scalewithsearch.com/articles/ai-agent-governance-guide-small-business) - [Migration and ownership](https://scalewithsearch.com/articles/leaving-vendor-ai-memory-guide) - Articles - [All articles](https://scalewithsearch.com/articles/) - [What changed](https://scalewithsearch.com/changes/) - About - Evidence - [Proof](https://scalewithsearch.com/proof) - Scale With Search - [About](https://scalewithsearch.com/about) - [How it works](https://scalewithsearch.com/how-it-works) # Recruiting and staffing AI rules: data, approvals, and spend Write staffing AI rules around the decision each function can affect. Preparing interview options, ranking candidates, and rejecting applicants are different activities. Give each activity its own review before choosing a tool. Start with a register of proposed functions. Name the business owner, records received, outputs produced, recipients, approval point, and spending limit. Keep an unapproved function disabled. The following operating policy is a proposed starting structure. Its legal references identify issues for the agency's responsible reviewer. They do not decide applicability to an unnamed agency, role, or jurisdiction. ## Separate office assistance from employment decisions DOJ guidance explains that employers must avoid disability discrimination when using hiring technology, including technology supplied by another company. [DOJ AI hiring guidance, checked 2026.10.26](https://www.ada.gov/resources/ai-guidance/). Have the employment reviewer inspect what the proposed tool does. A tool described as administrative may still affect opportunities if its output changes which applicants staff consider. For the first office build, prohibit candidate scores, rankings, inferred suitability, and automatic rejection. Allow only the named preparation task. Test that prohibition with requests to identify the best applicant. Require a new review if staff begin using summaries as selection advice. Record actual use in the function register. A written label cannot settle how people rely on an output. The [staffing first-functions plan](/articles/ai-native-staffing-first-functions) gives three narrower starting points. None authorizes expanding into candidate evaluation merely because the same tool could attempt it. ## Identify jurisdiction before enabling evaluation New York City's published AEDT guidance describes requirements for covered employers and employment agencies. These include a recent bias audit, public audit information, and notices before covered use. [NYC DCWP guidance, checked 2026.10.26](https://www.nyc.gov/site/dca/about/automated-employment-decision-tools.page). Do not treat that page as a nationwide rule or assume every office assistant is a covered tool. Ask the reviewer to determine whether the proposed use meets the applicable definitions. Record the work location, affected role, employer, agency responsibility, and actual tool behavior. Keep the determination with its source date. Reopen it when the function, geography, or use changes. A vendor's sales statement should enter the evidence file as a claim to check. The agency owner needs an explicit decision about its intended use before staff rely on the output. ## Keep restricted records out of general drafting As an agency operating rule, exclude Social Security numbers, bank details, identity documents, passwords, medical records, and background reports from general AI drafting. Keep those records in approved restricted systems. FTC guidance supports collecting and retaining only necessary sensitive information. [FTC data security guidance, checked 2026.10.26](https://www.ftc.gov/business-guidance/resources/protecting-personal-information-guide-business). A scheduling draft should use an assignment identifier, approved participants, and offered times. It should not receive a complete onboarding folder because the folder happens to be available. Separate accommodation requests from ordinary scheduling notes. Route the request to the designated reviewer. Give the drafting function only the approved operational instruction needed for the next step. Removing a name does not make every document suitable for general processing. Review free text, employer details, contact fields, and unusual circumstances before approving an extract. Keep the original evidence in the agency's controlled record. Use a documented selection process for the extract. Save the field list rather than asking each recruiter to improvise redaction. ## Define the data boundary for each function Use a small permission table that the owner can inspect. This is a proposed agency policy, not a claim that all listed records share one legal status. | Function | Approved input proposal | Excluded input proposal | Reviewer | | --- | --- | --- | --- | | Assignment change sheet | Accepted brief and relevant client request | Unrelated candidate records | Account manager | | Interview reply | Approved options and participant references | Medical explanation and payroll file | Coordinator | | Client update | Approved assignment events | Other clients and private assessments | Account manager | For each approved input, record where processing occurs and who can access it. Check the actual account settings and contractual terms before real records enter the function. Review retention, deletion, secondary use, subprocessors, and administrative access with the responsible data owner. Leave unresolved terms visible in the approval record. Do not translate unknown terms into a claim of privacy. Test with synthetic assignments first. Use obvious fake details so reviewers can practice spotting cross-client contamination without distributing a real person's record. ## Approve every outward message at the destination Require staff approval for candidate emails, client updates, interview invitations, job publications, and changes to public assignments. Bind approval to the final text and intended destination. An approved interview draft should not authorize a changed recipient or newly invented time. If either changes, return the revised message to review. Keep the accepted version with the assignment record. For confidential assignments, compare every public output with the approved disclosure boundary. The agency's [confidential job posting checklist](/articles/staffing-agency-confidential-job-postings) covers the website side of that handoff. Keep assignment acceptance, fees, candidate recommendations, offer terms, and start-date commitments with authorized people. Build the output queue so staff can see which decisions remain open. ## Make exception handling part of the policy Define a hold for missing assignment identity, conflicting times, uncertain recipient, unsupported status, and unexpected restricted data. Require a named person to resolve each hold. The function should save the reason and stop. It should not search more files until it finds a plausible answer. Give staff a route to complete the task manually. Treat instructions embedded in an applicant document as document content. They must not grant permission to reveal another record or change the agency's operating rules. Have the test reviewer insert such instructions into a synthetic attachment. Verify that the function preserves the boundary and reports the unexpected request for review. When a person overrides a hold, record the source and decision. One exception should not silently become a permanent agency instruction. Approve any standing rule change separately. ## Control spending before the queue runs Give each function its own monthly ceiling. Separate assignment preparation, scheduling drafts, and client updates in the cost record. Name the owner who can increase each limit. Set an alert before the ceiling and stop new model calls when the remaining approved budget cannot cover the next run. Reserve for retries and any separately billed processing. Use a bounded retry count. A missing candidate reference needs correction from staff. It should not consume repeated calls attempting to infer the person from partial details. Require a cost line even for failed runs. Keep the function identifier, assignment reference, model used, attempt count, outcome, and reviewer state together in the receipt. Where final billed cost arrives later, mark the first amount as provisional. Reconcile it against the provider's usage record. Do not label an estimate as a settled charge. Keep model identifiers in the private operating record. Replacement should require the same acceptance cases and data review. A lower unit price does not approve different processing terms. ## Give staff a policy they can run Use the 10/80/10 responsibility pattern: a person directs, the function prepares, and a person checks. Apply it to a named task and current source record. Train the account manager to inspect evidence behind a status statement. Train the coordinator to resolve a scheduling hold. Train the owner to locate receipts and stop a function. Review ordinary drafts, corrected drafts, holds, and cost exceptions together. Update the relevant source rule when a correction should affect future assignments. Keep the original decision visible. The [recruiting and staffing overview](/for/recruiting-and-staffing) connects these controls to agency office work. Bring Scale With Search one bounded task for an [AI systems scope](/ai-systems) through the [brief form](/work). Continue the website review: [AI-native staffing: the first functions to build](/articles/ai-native-staffing-first-functions); [Staffing application forms: help candidates finish](/articles/staffing-agency-application-form-errors); [Confidential job postings on a staffing agency website](/articles/staffing-agency-confidential-job-postings); [Staffing agency employer pages: turn requests into briefs](/articles/staffing-agency-employer-landing-page); [Expired staffing jobs: fix the website closure process](/articles/staffing-agency-expired-job-website-cleanup); [Staffing job alerts: define what candidates subscribe to](/articles/staffing-agency-job-alert-signup-page); [Staffing jobs missing from Google: inspect the detail page](/articles/staffing-agency-job-pages-google-discovery); [Staffing job salary fields: match the employer source](/articles/staffing-agency-job-salary-website-fields); [Local SEO for staffing agencies: match the office](/articles/staffing-agency-local-office-seo); [Recruitment scam warnings on staffing agency websites](/articles/staffing-agency-recruitment-scam-warning-page); [Remote staffing jobs: align locations and Google markup](/articles/staffing-agency-remote-job-location-markup); [Recruitment guarantees: explain the website conditions](/articles/staffing-agency-replacement-guarantee-page). [recruiting and staffing agencies](/for/recruiting-and-staffing): website and search resources. ## Related: recruiting and staffing agencies - [AI-native staffing: the first functions to build](/articles/ai-native-staffing-first-functions) - [Staffing application forms: help candidates finish](/articles/staffing-agency-application-form-errors) - [Confidential job postings on a staffing agency website](/articles/staffing-agency-confidential-job-postings) ---- ```text .|########||. .|########||. .|########||. |##||. .||##|. |##||. .||##|. |##||. .||##|. |#|. .|#|. |#|. .|#|. |#|. .|#|. |#| |#| |#| |#| |#| |#| .#| |#. .#| |#. .#| |#. .#. .#| .#. .#| .#. .#| |#. .#| |#. .#| |#. .#| |# ...... #| |# ...... #| |# ...... #| .# ||#########| #| .# ||#########| #| .# ||#########| #| .#. |######||######|. .#. .#. |######||######|. .#. .#. |######||######|. .#. #. .##|###|##|#|######| .# #. .##|###|##|#|######| .# #. .##|###|##|#|######| .# || |##|#||||||||||||#||#| || || |##|#||||||||||||#||#| || || |##|#||||||||||||#||#| || # |#||||||||||||||||||||#| #. # |#||||||||||||||||||||#| #. # |#||||||||||||||||||||#| #. || |#||||||||||||||||||||||#| || || |#||||||||||||||||||||||#| || || |#||||||||||||||||||||||#| || # .#||||||||||||||||||||||||#| # # .#||||||||||||||||||||||||#| # # .#||||||||||||||||||||||||#| # || ....|||#||||||##|#|||#|#||##||||....|. || || ....|||#||||||##|#|||#|#||##||||....|. || || ....|||#||||||##|#|||#|#||##||||....|. || #. . ....|# ....#||| ||| .#||| ....#. .# #. . ....|# ....#||| ||| .#||| ....#. .# #. . ....|# ....#||| ||| .#||| ....#. .# # . ||||#| .#####|| . .#| .#||| ||||# #. # . ||||#| .#####|| . .#| .#||| ||||# #. # . ||||#| .#####|| . .#| .#||| ||||# #. .| ||||| #. |#|||||. || #. |#||. .||||| || .| ||||| #. |#|||||. || #. |#||. .||||| || .| ||||| #. |#|||||. || #. |#||. .||||| || || |.... #. ....|#. |. ...|. ....|| || || |.... #. ....|#. |. ...|. ....|| || || |.... #. ....|#. |. ...|. ....|| || #. .||||||##||||||##||####|||||||#||||||#| .# #. .||||||##||||||##||####|||||||#||||||#| .# #. .||||||##||||||##||####|||||||#||||||#| .# # .||####|###########################|. # # .||####|###########################|. # # .||####|###########################|. # # .#||||||.#.|| # |. ..# #| #|||||#| # # .#||||||.#.|| # |. ..# #| #|||||#| # # .#||||||.#.|| # |. ..# #| #|||||#| # # .#|||||| .. |# ## |#| ...#|#|||#| # # .#|||||| .. |# ## |#| ...#|#|||#| # # .#|||||| .. |# ## |#| ...#|#|||#| # # .#|||#|# .# .#| #| ##|.#..#|||||#| # # .#|||#|# .# .#| #| ##|.#..#|||||#| # # .#|||#|# .# .#| #| ##|.#..#|||||#| # # .#||||||############|######|||||#| # # .#||||||############|######|||||#| # # .#||||||############|######|||||#| # # |...||#||||||#||||#|#||||||#|||||#||| |.| # # |...||#||||||#||||#|#||||||#|||||#||| |.| # # |...||#||||||#||||#|#||||||#|||||#||| |.| # #. .. ||||# .|||##|. |#| .|| || .|||#. #| # .# #. .. ||||# .|||##|. |#| .|| || .|||#. #| # .# #. .. ||||# .|||##|. |#| .|| || .|||#. #| # .# || | ...|| ...##| | #| .|. |. ##### .. .| || || | ...|| ...##| | #| .|. |. ##### .. .| || || | ...|| ...##| | #| .|. |. ##### .. .| || || ||||| || ||||#| . |. |. |#. ||||| |#| || || || ||||| || ||||#| . |. |. |#. ||||| |#| || || || ||||| || ||||#| . |. |. |#. ||||| |#| || || .# |.....#|....|#.||||.|||##.|#|....||.#||.#. #. .# |.....#|....|#.||||.|||##.|#|....||.#||.#. #. .# |.....#|....|#.||||.|||##.|#|....||.#||.#. #. #.|||||||#############################| ||| .# #.|||||||#############################| ||| .# #.|||||||#############################| ||| .# || ##||#||#||||||#||#||#||#||##. || || ##||#||#||||||#||#||#||#||##. || || ##||#||#||||||#||#||#||#||##. || # .#|||||||||#||#|||||||||||#| # # .#|||||||||#||#|||||||||||#| # # .#|||||||||#||#|||||||||||#| # || |#||||||||||||||||||||||#| || || |#||||||||||||||||||||||#| || || |#||||||||||||||||||||||#| || .# |#||||||||||||||||||||#| #. .# |#||||||||||||||||||||#| #. .# |#||||||||||||||||||||#| #. || |#||||||||||||||||||#| || || |#||||||||||||||||||#| || || |#||||||||||||||||||#| || #. .######|#|#########| .# #. .######|#|#########| .# #. .######|#|#########| .# .#. |#####||||#####| .#. .#. |#####||||#####| .#. .#. |#####||||#####| .#. |# ||########|| #. |# ||########|| #. |# ||########|| #. |# ...... #| |# ...... #| |# ...... #| |#. .#| |#. .#| |#. .#| |#. .#. |#. .#. |#. .#. .#| |#. .#| |#. .#| |#. |#| |#| |#| |#| |#| |#| .|#|. .|#|. .|#|. .|#|. .|#|. .|#|. |##||. .||##| |##||. .||##| |##||. .||##| .||########||. .||########||. .||########||. Scale With Search 2026 [scalewithsearch.com](https://scalewithsearch.com) ```