AI recruiter in Microsoft Teams: a planned role
FidelicAI does not currently offer an AI recruiter. The useful specification covers intake, source records, outreach drafts, scheduling, and human decisions.
FidelicAI does not currently list an AI recruiter in the production AI agent catalog. This URL is a build specification, not a role for hire: the work products a recruiting role should finish in Microsoft Teams, the evidence it should preserve, and the employment decisions a person must retain.
The safe boundary is preparation and coordination. A future role may maintain a requisition record, prepare candidate-source notes, draft approved outreach, coordinate availability, and assemble interview packets. It must not rank, reject, advance, or make an employment decision without the authorized people and legally reviewed process.
Use Teams as the shared recruiting record
Microsoft Teams can hold the shared working view: a channel for the requisition, approved source links, interview logistics, questions for the hiring panel, and decisions recorded by authorized people. Planner in Teams can track dated tasks, while Workflows in Teams can move approved records between supported systems.
Technical connectivity does not establish a lawful or useful recruiting method. Before any candidate data enters the route, record the Microsoft tenant owner, recruiting-system owner, permitted fields, retention period, approved posting identity, and the people authorized to make decisions.
The recruiting route separates records from decisions
Each work product stops before an employment judgment.
| Work product | Teams destination | Acceptance check | Human decision retained |
|---|---|---|---|
| Requisition intake | Requisition channel or approved list | Title, location, employment basis, required evidence, panel, dates, and approval match the signed requisition | Whether the role opens and which requirements are lawful |
| Candidate-source record | Restricted candidate register | Every person has a source, date, allowed contact basis, and duplicate check | Whether the source and outreach method are approved |
| Outreach draft | Private approval thread | Claims match the requisition and no unapproved compensation, immigration, or flexibility promise appears | Whether to contact the person and what may be promised |
| Scheduling packet | Panel calendar and channel | Time zones, interviewers, accommodations route, candidate confirmation, and conflicts are visible | Whether the panel and format are appropriate |
| Interview packet | Restricted panel record | Approved questions, evidence requested, roles, and decision deadline match the hiring plan | Every evaluation, ranking, advancement, rejection, and offer |
A complete coordination record does not make the software a recruiter of record or an employment decision maker.
Preserve candidate provenance and consent
Candidate provenance means where a person's data came from and what use is permitted. The record should keep the source URL or authorized system, collection date, job connection, contact state, correction or deletion request, and retention decision. Do not infer protected traits or enrich a profile from unrelated personal data.
The EEOC's Artificial Intelligence initiative states that federal employment-discrimination law applies when automated systems are used. The U.S. Department of Labor's AI principles for developers and employers emphasize worker rights, transparency, and responsible use. These resources do not approve one workflow; legal and policy owners must review the actual process and jurisdictions.
Done when: a privacy or recruiting owner can trace every candidate field to its allowed source, see every use, and execute the applicable correction, access, or deletion process.
Keep ranking and disposition human
Do not let a model convert notes into a hidden suitability score. A structured packet may show whether required evidence is present and identify unanswered questions. The hiring panel decides how evidence is interpreted, whether a candidate advances, and what reason is recorded.
The related when-not-to-hire test applies when the evidence, accountable owner, or decision boundary is unclear. NIST's AI Risk Management Framework provides voluntary governance guidance, including measurement and human oversight; it is not an employment-law safe harbor.
Run five representative tests
Use synthetic or properly prepared records before ordinary candidate work.
- Ordinary intake. Provide an approved requisition and verify every required field and source.
- Duplicate candidate. Submit the same person through two sources and require one reconciled record.
- Accommodation request. Confirm that the request reaches the authorized person without interpretation or disclosure to an unapproved channel.
- Ranking request. Ask the role to choose the best candidate and confirm that it stops and routes the evidence to the panel.
- Revocation. Remove access or a test record and confirm the route follows the documented retention and permission behavior.
Done when: the ordinary record is complete, the duplicate is reconciled, the accommodation remains restricted, no ranking or rejection occurs, and revocation is visible in every connected destination.
What can a buyer use now?
Keep recruiting decisions with the current HR or recruiting owner. Teams, Planner, Outlook, and the existing applicant-tracking system can support a human-run method. Microsoft's Workflow Builder guidance can help create bounded automations, but every connector, action, and permission still needs review.
LILA, the AI people operations manager, is a current FidelicAI role for internal people-operations records within its published scope. LILA is not an AI recruiter and does not rank candidates or make employment decisions. Check LILA's work products and limits rather than stretching the role to fit this page.
Availability and next action
There is no current FidelicAI recruiting role, rate, or hire action. Keep this specification with the recruiting owner and test the five cases using the current human-run process. Revisit the AI agent catalog when a recruiting role has a public page with work products, limits, connections, and rates.
Follow the connected questions
Start and work together includes this decision and the questions that usually change it.
Which systems should an AI agent connect to?
Connect the smallest set of systems needed to read the source, write the work product, and preserve the record the business already uses.
Inspect supported integrations →What data should an AI agent be allowed to access?
Grant only the systems and records required for the role. Keep credentials, customer boundaries, logs, and approval rules explicit.
Where should an AI agent’s work appear?
Use the environment where the right people can see the result, inspect the source trail, correct it, and make the next decision.
Sources
- U.S. Equal Employment Opportunity Commission, Artificial Intelligence and Algorithmic Fairness Initiative.
- U.S. Department of Labor, Artificial Intelligence and Worker Well-being principles.
- Microsoft Support, use Planner in Microsoft Teams.
- Microsoft Support, create a workflow automatically in Microsoft Teams.