SolutionPath — AI Book Generation Platform & Google Workspace Integration

Overview & Greenfield Architecture

SolutionPath is an automated publishing SaaS designed to transform structured chapter outlines into complete, multi-language books. I built this application from scratch as a greenfield project, taking responsibility for the full-stack architecture, background worker orchestration, database design, and external cloud integrations.

The core frontend was built as a React and Vite single-page application using TypeScript, Zod, and TanStack Query. On the backend, I set up FastAPI alongside PostgreSQL, using SQLAlchemy and Alembic to handle schema design and structured database migrations.

End-to-End Generation & Translation Workflow

The platform coordinates several services to process long-form content reliably without running into HTTP request timeouts.

Users start by logging in with their Google account and submitting a link to a Google Sheet from their Drive. The backend parses this Sheet dynamically to extract chapter titles, summaries, and structural metadata. Through the frontend interface, users can select which chapters to generate, choose writer personas, adjust target chapter lengths, and select anywhere from one to eight target languages.

Once generation begins, the outline and configuration are stored in PostgreSQL. The heavy processing is immediately offloaded to Celery background workers connected via Redis. The system generates the book chapter by chapter in the primary language first, then kicks off parallel asynchronous translation jobs across all selected secondary languages. Throughout this multi-step process, the frontend streams live status updates so users can track generation and translation progress in real time. Once finished, users can export the completed, formatted books directly as new Google Docs in their connected Google Drive accounts.

Navigating the Google OAuth Verification Process

While the core application logic and AI pipelines were completed in about a month and a half, the project spanned four months due to the strict Google Cloud OAuth verification process.

Because the application needed to read user-provided Google Sheets and write newly generated Google Docs, it required sensitive and restricted Google API scopes. Google's Trust & Safety team repeatedly pushed to narrow our permissions down to the restrictive drive.file scope. While drive.file worked cleanly for creating and exporting new Docs, it completely blocked the backend from reading existing Google Sheets that users linked from their own Drive.

To resolve this bottleneck, I spent weeks testing alternative permission boundaries, preparing technical compliance documentation, recording detailed video walkthroughs of the user flow, and explaining our architectural necessity directly to Google's review team. Through persistent revisions and justification, I successfully secured full production OAuth approval.

Deployment & Production Delivery

After obtaining official Google Cloud verification, I deployed the complete system onto the client's Linux VPS. I containerized the Vite frontend, FastAPI backend, Celery workers, and Redis broker, configuring Nginx reverse proxies, SSL certificates, and DNS settings to bring the platform into active production.