> ## Documentation Index
> Fetch the complete documentation index at: https://irisdocs.prescientlabs.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Worked Example — Recruitment Firm Workflow

> Illustrative diagnostic for a recruitment firm experiencing manual candidate, client and CRM administration.

<Note>
  This is an illustrative training example, not a named client engagement.
</Note>

## Situation

A 90-person recruitment firm has grown quickly. Consultants spend substantial time after calls updating CRM records, writing candidate summaries and coordinating interview stages.

The managing director believes the answer is “an AI recruiter.”

## Discovery

The consultant asks where revenue-producing staff lose time and which delays affect candidate or client experience.

The most consequential issues are:

* consultants re-enter call notes into the CRM;
* candidate summaries vary significantly by consultant;
* interview coordination creates repeated back-and-forth;
* recruiters search old notes manually before client calls;
* managers do not trust pipeline data because updates are inconsistent.

## Diagnostic scope

**Business question:** which activities from candidate/client conversation through CRM update and next-action coordination create avoidable consultant effort or poor data quality?

The scope deliberately excludes automated candidate-selection decisions.

## Stakeholders

* managing director;
* recruitment team lead;
* three recruiters from different desks;
* operations/CRM owner;
* one compliance representative.

## Evidence

Stakeholder interviews show major differences in working style. Screen observation with consent reveals that some recruiters use personal notes, email and CRM in parallel. CRM extracts quantify missing fields, timing of updates and duplicate records.

## Findings

1. The workflow has no standard “conversation-to-record” step.
2. Recruiters recreate information already present in call notes and email.
3. Data quality problems are partly caused by poor workflow design, not recruiter resistance.
4. Candidate summaries are high-volume but still require recruiter judgment and client context.
5. Interview scheduling is rules-heavy and integration-friendly.

## Opportunities

### Conversation-to-CRM assistance

Create structured first-draft CRM updates from approved conversation notes, with recruiter review.

### Candidate-summary drafting

Generate a first draft using approved candidate information and role context; recruiter owns final interpretation.

### Scheduling workflow

Use deterministic scheduling/integration before adding AI complexity.

### Knowledge retrieval

Improve access to historic client/candidate context from approved internal records.

## Prioritisation

The consultant recommends conversation-to-CRM first because it combines:

* high frequency;
* visible consultant pain;
* clear human review;
* measurable data-quality improvement;
* low dependence on high-impact automated decision-making.

Scheduling automation follows as a second contained workstream.

## Business case

The model uses:

* calls per consultant per week;
* current admin minutes per call;
* percentage of records requiring later correction;
* loaded consultant cost;
* expected adoption range.

The value case includes capacity for additional candidate/client activity but does not automatically convert every hour into profit.

## Executive recommendation

> Do not begin with autonomous candidate selection. Start by reducing the administrative work around conversations and scheduling, where the evidence is strongest and the control model is straightforward.

## Expansion

If data quality improves, later work could examine pipeline forecasting, internal knowledge retrieval and client-report preparation.