Autonomous AI-driven learning platform that transforms learner onboarding, planning, content curation, assessment, and motivation into one intelligent experience.
Built by Team Antigravity, SkillCrew is a full-stack, multi-agent scaffold designed to move beyond static courses and isolated chatbots, creating a continuous, adaptive learning loop.
SkillCrew is a modern learning ecosystem powered by five specialized autonomous agents. It combines a high-concurrency FastAPI backend with a reactive Next.js 16 frontend to automate the entire educational lifecycle:
- Intelligent Onboarding: Extracting intent and experience from resumes and LinkedIn profiles.
- Adaptive Architecting: Building and revising multi-week roadmaps based on real-time performance.
- Smart Curation: Discovering high-quality resources (YouTube, Coursera, etc.) tailored to the learner's pace.
- Active Retention: Implementing spaced-repetition and automated memory reinforcement workflows.
- Motivational Nudges: Multi-channel engagement via WhatsApp, SMS, and Email to maintain momentum.
The core of SkillCrew is built around five distinct AI personas, each responsible for a specific domain of the learner journey:
| Agent | Persona | Responsibility |
|---|---|---|
| Nova | Intake Specialist | Parses resumes/LinkedIn to create the canonical learner profile. |
| Archie | The Architect | Designs and dynamically revises milestone-based learning roadmaps. |
| Dexter | Resource Curator | Discovers and ranks learning resources from across the web. |
| Pip | Memory Guardian | Generates active recall quizzes and spaced-repetition revision packs. |
| Sparky | Motivation Coach | Sends motivational nudges and gamified alerts via Twilio/Resend. |
- Unified End-to-End Flow: Seamlessly transitions from onboarding to assessment and roadmap adaptation.
- Profile-Aware Personalization: Bridges the specific gap between your professional history and target role.
- Domain-Agnostic Planning: Structures milestones for any field—from software engineering to digital marketing.
- Adaptive Feedback Loop: If a quiz shows weakness, Archie automatically revises upcoming modules.
- Integrated Assessment: Quizzes aren't just for grades; they are signals that feed back into the learning logic.
- Framework: Next.js 16 (App Router), React 19
- Styling: Tailwind CSS, Lucide React
- Animations: Framer Motion (Scroll-linked "scrollytelling")
- State/Auth: Supabase, Zod
- Framework: FastAPI, Uvicorn, Pydantic
- LLMs: Google Gemini (1.5 Pro/Flash), Groq (Llama 3)
- Search & Ingestion: Apify (LinkedIn Scraping), Tavily Search API
- Communications: Twilio (WhatsApp/SMS), Resend / SendGrid (Email)
- Media: YT-DLP, YouTube-Transcript-API
graph TD
A[User/Resume] -->|Nova| B(Learner Profile)
B -->|Archie| C(Dynamic Roadmap)
C -->|Dexter| D(Curated Content)
D -->|Pip| E(Assessment/Retention)
E -->|Feedback Loop| C
E -->|Sparky| F[WhatsApp/Email Notifications]