AI Agent Center
Six agents working in the background. Placeholder logic today, swap in Claude or OpenAI when ready.
Student Roadmap Agent
placeholderGuide each student through the 13 PPL milestones.
Capabilities
- • Explain the current step in plain English
- • Recommend the next action
- • Identify blockers (missing medical, IACRA, endorsements)
- • Generate per-step checklists
Functions
- analyzeStudentRoadmap()
- recommendNextRoadmapAction()
- explainRoadmapStep()
- detectRoadmapBlockers()
Instructor Admin Agent
placeholderReduce CFI admin and repetitive explanations.
Capabilities
- • Generate lesson debriefs from quick notes
- • Summarize student progress
- • Draft instructor messages
- • Identify students needing attention
Functions
- generateLessonDebrief()
- summarizeStudentProgress()
- draftInstructorMessage()
- identifyStudentsNeedingAttention()
- recommendNextLesson()
Checkride Prep Agent
placeholderRun mock orals and audit checkride readiness.
Capabilities
- • Run AI-DPE mock orals with follow-up questions
- • Score answers and identify weak ACS areas
- • Generate checkride readiness reports
- • Build per-student prep plans
Functions
- startMockOral()
- evaluateOralAnswer()
- generateFollowUpQuestion()
- calculateCheckrideReadiness()
- generateCheckridePrepPlan()
Document Readiness Agent
placeholderMake sure checkride paperwork is bulletproof.
Capabilities
- • Detect missing documents
- • Flag unverified items
- • Generate the DPE document checklist
- • Alert instructors to missing endorsements
Functions
- analyzeDocumentReadiness()
- generateDocumentChecklist()
- detectMissingEndorsements()
- flagCheckridePaperworkIssues()
Scheduling Agent
placeholderKeep training cadence and the calendar honest.
Capabilities
- • Suggest next lesson slots
- • Detect long gaps in training
- • Recommend study blocks
- • Build checkride-prep schedules
Functions
- suggestNextLessonSlots()
- detectTrainingGaps()
- createStudySchedule()
- createCheckridePrepSchedule()
- recommendCalendarEvents()
Wiring it up
Every function above lives in lib/ai/agents.ts with stable signatures. Replace each function body with a Claude or OpenAI call (with prompt caching enabled) and the rest of the app keeps working without changes. The calendar service uses the same pattern in lib/calendar/calendarService.ts.