Guide: generate a schedule with Athena AI

Orkelis includes an artificial intelligence engine called Athena, capable of automatically generating your schedules while respecting legal constraints, employee contracts, and your staffing needs. This feature transforms hours of manual work into seconds of automated computation. This guide details each step, from initial setup to accepting the generated schedule.

Prerequisites

Before using automatic generation, ensure the following conditions are met:

• Active Pro or Enterprise plan: AI is subject to a monthly token quota (included in Pro, unlimited in Enterprise).
• At least one service with assigned employees: the engine needs available employees to create the schedule.
• Shift types configured in Settings > Shift Types (e.g., Morning 6 AM - 2 PM, Afternoon 2 PM - 10 PM, Night 10 PM - 6 AM).
• Contractual hours filled in for each employee: Athena respects the hour volumes defined in profiles.
• Specialties up to date: if your positions require specific qualifications (e.g., only an RN can be assigned to care), specialties must be configured.

Preparation: the more complete and up-to-date your data is (contractual hours, specialties, preferences), the better the automatic generation result will be. Take 10 minutes to verify your employees' profiles before your first generation.

1. Launch an automatic generation

Automatic generation creates a complete schedule for a given service and week, taking all configured constraints into account.

1. Go to the Schedule page (/dashboard/planning) from the sidebar.
2. Select the relevant service from the dropdown at the top of the page.
3. Navigate to the desired week using the navigation arrows.
4. Click the Generate with AI button (purple lightning icon) in the action bar.
5. A modal window opens with generation options.
6. Choose the optimization mode based on your priority:

• Balanced: distributes shifts and hours fairly among all employees. Ideal for teams where fairness is important.
• Coverage: maximizes coverage of priority time slots, even if it creates imbalances between employees. Ideal during high-activity periods.

1. Optional: add manual constraints in the text field. Examples: "Marie does not work on Mondays", "Minimum 3 RNs in the morning", "Paul off on Wednesday and Thursday".
2. Optional: filter by specialty if you only want to generate the schedule for a subset of employees.
3. Click Launch generation.
4. Generation typically takes between 5 and 30 seconds depending on team size. A progress bar shows you the status.

Tip: if you have recurring weekly constraints (e.g., an employee on half-time only on Mondays and Tuesdays), configure them in the employee's availability from their profile. Athena will automatically take them into account without you having to enter them each generation.

2. Understanding the constraints considered

The Athena engine considers a comprehensive set of constraints to generate a legal and operational schedule:

Automatic legal constraints

• Minimum daily rest of 11 hours between two consecutive shifts.
• Minimum weekly rest of 35 consecutive hours (24h + 11h).
• Maximum daily working time (10 hours, except by derogation).
• Maximum weekly working time (48 absolute hours, 44 hours average over 12 weeks).
• No work for more than 6 consecutive days.

Contractual constraints

• Respect for each employee's contractual weekly hour volume.
• Consideration of contract type (permanent, fixed-term, part-time, temporary).
• Respect for availability days configured in the employee's profile.

Operational constraints

• Staffing needs per slot (minimum number of employees required per shift type).
• Required specialties (e.g., at least 1 RN and 1 NA in the morning).
• Approved absences (employees on leave, sick leave, or training are automatically excluded).
• Manual constraints entered by the manager.

3. Multi-week generation

To plan over several consecutive weeks (useful for monthly schedules or rotations):

1. Use the same Generate with AI button.
2. In the configuration window, enable the Multi-week option.
3. Select the date range: start date (Monday of the first week) and end date (Sunday of the last week).
4. Select the optimization mode.
5. Click Launch generation.

The AI accounts for continuity between weeks to avoid overlap issues, respect weekly rest, and balance hours across the entire period. Multi-week generation consumes approximately 5,000 tokens per generation.

Rotations: if you use a rotation system (e.g., morning week / afternoon week), configure the cycles in Settings > Rotations. Athena will automatically respect them during multi-week generation.

4. Natural language prompt generation

For users who prefer expressing their needs in plain language rather than checking options, Orkelis offers a conversational AI assistant:

1. Open the AI chat by clicking the bubble at the bottom right of the screen or the Athena button in the action bar.
2. Type your request in natural language. Examples:

• "Generate a schedule for the week of March 10 with Marie off on Monday and Tuesday."
• "I want 3 people in the morning and 2 in the afternoon every day."
• "Schedule Sophie only on night shifts this week."
• "Create a balanced schedule but make sure Paul and Jean never work the same day."
• "Generate the March schedule for the Emergency department."

1. The AI parses your request and displays a summary of extracted constraints for validation.
2. Review the proposed constraints. You can modify or add to them.
3. Click Generate to launch the computation.
4. The generated schedule appears in draft mode for your validation.

Prompt history: all your prompts are saved automatically. Find them in the Prompt History menu to reuse or adapt a formulation that worked well.

5. Review and accept the result

Once generation is complete, the proposed schedule appears in draft mode (dotted background). No employee is notified until you accept and publish the schedule.

1. Check the health score displayed in a badge at the top of the schedule:

• Green score (90-100%): optimal schedule, all constraints respected.
• Orange score (70-89%): acceptable schedule with minor compromises.
• Red score (< 70%): significant issues, modifications recommended.

1. Review any alerts in the right side panel: understaffing on a slot, hour overrun for an employee, rest not respected.
2. For each alert, Athena suggests a correction. Click Apply to accept the suggestion or modify manually.
3. Adjust the schedule manually if needed: drag and drop a shift to move it, click a cell to modify or delete a shift.
4. Once satisfied, click Accept to validate the draft.
5. Then click Publish the week to notify your employees.
6. If the result doesn't suit you, click Cancel to delete the draft and start over.

6. Contextual suggestions

Beyond full generation, the AI proactively offers suggestions whenever you edit the schedule:

• Understaffing alerts: if a slot doesn't have enough employees, an orange banner appears on the cell with the missing count.
• Hour overrun detection: when an employee approaches their weekly limit, their hour total turns orange then red.
• Replacement proposals: when an absence is approved, Athena automatically identifies available and qualified employees to replace. A Replace button appears next to the absence.
• Energy optimization: Athena detects employees working too many consecutive days and suggests swaps.

These suggestions appear in the right side panel of the schedule, accessible by clicking the lightbulb icon. Each suggestion can be applied in one click or dismissed.

Pro tip: run an Athena check after each manual schedule modification by clicking the Verify button (check + lightning icon). This updates the health score and alerts in real time.

7. Limits and quotas

AI engine usage is subject to a monthly quota system to ensure optimal service quality:

• Pro plan: monthly AI token quota (renewed on the 1st of each month). Each simple generation consumes approximately 2,000 tokens. Each multi-week generation consumes approximately 5,000 tokens.
• Enterprise plan: unlimited AI engine usage, with no restrictions.
• Contextual suggestions and verifications consume only a negligible number of tokens.

Check your token consumption in Settings > AI Usage. If your quota is reached, you can upgrade to the Enterprise plan or wait for the automatic monthly renewal.

8. AI generation troubleshooting

• Generation fails with "Impossible constraints": your constraints are contradictory (e.g., too few employees to cover all slots). Reduce staffing needs or add employees to the service.
• Health score is very low: verify that your employees' contractual hours are correct and that you have enough employees for the service.
• Generation is very slow (> 1 minute): this happens when the team is very large (> 50 employees) or the multi-week range is very long. Try generating in smaller groups.
• An employee is scheduled while on leave: verify that the absence is properly recorded and approved in Absences. Only absences with "Approved" status are considered by Athena.
• The Generate with AI button is grayed out: your plan doesn't support AI (Essential plan) or your monthly quota is exhausted.
• Manual constraints are not respected: rephrase your constraint more explicitly. Example: instead of "Marie prefers mornings", write "Marie only on Morning shift".

AI support: if you encounter an unexpected result, contact our team via the support chat. Attach a screenshot of the generated schedule and your constraints for a quick diagnosis.