AI for Nanny Agencies: Placements, Screening and Schedules
How a nanny agency can use AI agents for placement tracking, screening status and schedule requests, and which background check decisions stay with staff.

An AI agent for a nanny agency can keep track of where every caregiver and family request stands in the placement pipeline, report screening status from records staff have approved, and handle routine schedule requests within limits the agency defines in advance. It should not decide whether a candidate is suitable, interpret a background check, or tell a family that a caregiver is cleared. Those judgments stay with trained staff, and the agent's job is to make sure they are made on time and recorded clearly.
Key takeaways
- Delegate placement tracking, reminders, document follow-up and schedule intake. Keep eligibility, safety and hiring judgments with people.
- Give the agent screening status fields rather than full reports, unless a documented policy requires more access.
- Write the approval boundaries before deployment so every schedule change that needs a coordinator is routed there automatically.
- Judge the agent by whether each open placement has a named next step and a date, not by how many messages it sends.
Why nanny agencies lose track of placements
Small and mid-sized nanny agencies often run their pipeline across email threads, shared spreadsheets, phone notes and a coordinator's memory. A caregiver may be waiting on a reference, a family may have changed a weekday schedule twice, and a trial start date may sit in a message nobody has reopened. None of these gaps is dramatic on its own. Together they cause missed follow-ups, repeated calls and placements that begin without a clear record of what was agreed.
An agent with access to a structured record of each placement can hold that context from one day to the next. It can tell the coordinator which placements have no next action, which documents are still missing and which schedule requests have gone unanswered. This only works if staff maintain the record and the agent's scope is written down.
What a placement tracking agent can handle
A practical first version tracks five kinds of information:
- Caregiver profile status, including documents received, documents missing and expiry dates that staff have entered
- Family request details, such as hours, location, languages spoken, children's ages and any needs staff have noted
- Candidate offers, responses and the date of each response
- Trial periods, start dates and scheduled check-ins after placement
- Issues raised after placement, each with an owner and a due date
Within that scope, the agent can draft a morning summary for the coordinator, flag records that have passed their next due date, and prepare a suggested message for the coordinator to approve. It should not send a placement confirmation or change a caregiver's assignment without a named person signing off. Agencies that already track credential expirations face a similar problem, which is covered in AI for a Credentialing Company: Tracking Applications and Expirations.
Screening status: what the agent reports and what it never decides
Screening is where the boundary matters most. Background checks, identity verification and reference reviews produce personal data that is sensitive for the candidate and carries legal obligations for the agency. Requirements differ by country and by state or region, and in some places consumer reporting rules govern how results are obtained and used. An agency should review its obligations with counsel before connecting any system to screening data.
The safe pattern is narrow. The agent reads a status label that staff set, such as requested, received, under review, cleared by staff or flagged for review. It can remind a candidate which items are still missing, using approved wording. It should not summarize a report, rank candidates by anything found in a check, or tell anyone that a candidate has passed.
Use this decision guide when you define the workflow:
- Sending a standard reminder for a missing document, using an approved template: the agent may do it
- Updating a status field after staff have recorded the result: the agent may do it
- Reading or summarizing the content of a screening report: staff only
- Deciding whether a flagged result affects eligibility: staff only, with the reason written down
- Telling a family that a caregiver is cleared to start: staff only
Illustrative scenario: a coordinator marks a candidate's background check as flagged for review. The agent moves the record into a review queue, notifies the named reviewer, and holds every message to the candidate and the family until the reviewer records a decision. It does not open the report, suggest an outcome or guess how long the review will take.
Handling schedule requests without double booking
Schedule requests arrive in many forms. A family asks to add a weekend, a caregiver asks to swap a shift, a parent wants a holiday covered. The agent's first job is to turn each message into a structured request with dates, times, change type and urgency, and to ask for anything missing. Its second job is to check that request against confirmed placements, the caregiver's recorded availability and the agency's rules on notice periods and travel time.
The agent can propose options. It can apply changes only where an approval boundary already allows it, for example a small move of a confirmed shift when the caregiver has confirmed availability in the record. Any change that touches pay, overtime, a caregiver's assignment or a family's contract goes to a coordinator. Every proposed and applied change should be logged with the request it came from. Parent communication raises similar questions about records and approvals, as described in Superintelligent Agents for Private Schools: Parent Messages and Records.
Messages from families and candidates can contain text that looks like an instruction, such as a request to ignore the approval rules. Treat that text as content to be read, never as a command. The OWASP guidance on risks in LLM applications describes this class of problem in detail.
Permissions, memory and the audit trail
Access should follow the principle of least privilege. The agent gets read access to what it needs and write access only to the fields it is allowed to update. If the agency connects a calendar or email account, the authorization should cover those functions rather than the whole account. OAuth 2.0 is the common framework for granting this kind of limited access, and OWASP's authorization guidance is a useful reference for checking that each action is permitted for the requester and the role.
Memory needs the same discipline. Keep a short, factual record of each placement, updated by staff and by logged agent actions, rather than letting the agent accumulate impressions about families or caregivers. Review that memory when a coordinator changes roles or when a placement ends, and remove anything that no longer serves the work. The NIST AI Risk Management Framework offers a sound structure for deciding which risks to measure and who owns each control.
Step-by-step plan for a first deployment
Start with one role, usually the coordinator's daily tracking work, and follow this sequence:
- Map the pipeline stages from first contact to post-placement check-in, and name the person who owns each stage.
- List every field the agent may read and every field it may write. Mark screening fields as read-only unless policy says otherwise.
- Write the approval boundaries for schedule changes and screening messages in plain language, then translate them into the workflow.
- Choose one success criterion staff can check, such as every open placement having a named next action and a due date.
- Run a review period in which the agent drafts messages and a coordinator approves or edits each one before anything is sent automatically.
- Review permissions and memory on a fixed schedule and after any staff change.
How Be Human approaches this as a managed service
Be Human designs this kind of setup as a managed service. A first engagement defines one useful role, the memory and tools that role needs, the approval boundaries for schedule changes and screening messages, and one observable success criterion. Managed operations can include monitoring of the agent's actions, recurring evaluations of drafted messages, permission reviews when staff roles change, incident response and controlled changes to the workflow. Service is remote, in English and Portuguese.
Measuring whether the agent is doing its job
Measure outcomes that staff can see in the record. Do open placements carry a next action and a date? Do screening reminders go out on the approved schedule? Can every schedule change be traced to a request and an approver? Those questions show whether the agent is reducing dropped follow-ups without taking over judgments it should not hold.
Avoid measuring by message volume. A busy agent that drafts many messages nobody reads has not helped the coordinator. Review a sample of logged actions regularly, and treat any action that fell outside its boundary as an incident to investigate, not a detail to smooth over. If you want to identify the first task in your agency worth handing to an agent, the free Be Human AI diagnostic is a practical place to start.
Frequently asked questions
Can an AI agent run background checks for a nanny agency? An agent can track whether a check has been requested, received or is under review, based on status that staff record. Reading the report, deciding eligibility and ordering any new check stay with staff, who must follow the rules that apply in their jurisdiction.
Does an AI agent replace a placement coordinator? No. It handles tracking, reminders, drafts and logging so the coordinator can spend more time on matching, judgment and family conversations. The coordinator remains the approver for any schedule change that affects pay, contracts or assignments.
Which schedule requests are safe to automate? Requests that fit a boundary the agency has already approved and that can be checked against confirmed records are the best candidates. Examples include collecting the details of a new request, proposing open time slots and making small adjustments when a caregiver's availability is confirmed in the record. Changes involving overtime, caregiver assignment or a family's contract should go to a person.
How should an agency plan the timeline for a first deployment? Plan by milestones rather than by a fixed date, because the timeline depends on how complete the existing records are and how many systems are involved. A focused first role with clear boundaries is easier to review and correct than a broad rollout across the whole agency.
Your next step
See where an AI agent fits in your company first.
Take the free Be Human AI diagnostic. A few questions about your work show which task to hand to an agent first, what it needs to know and where a person approves.
Be Human editorial guidance. Examples are illustrative unless explicitly attributed. Sources provide background; implementation choices depend on your workflow.


