Campus Dyno CPQ

What it is

The CPQ platform I wished every deal desk I ran had. Multi-tenant from day one, built on a TypeScript monorepo (Express 5, PostgreSQL with Drizzle, React), and running in production.

Quote lifecycle

• Tenant-isolated magic-link auth
• Deterministic pricing engine: net price, discounts, ramps, totals
• Discount-threshold approval routing: auto, manager, or VP based on depth
• Framework agreements with pre-approved discount ceilings
• Branded PDF and Google Doc generation with automatic template selection
• Per-tenant catalog spanning per-seat, usage-based, platform, and one-time pricing, resolved through a 6-level SKU-resolution waterfall

Document ingestion

Upload a prospect's order form as PDF, DOCX, or image and get a templated quote back:

• PDF parsing with line detection, DOCX extraction, and vision OCR for images
• Field-to-schema mapping against a slug registry, with deterministic alias resolution before any LLM call and a validation guard that strips hallucinated fields
• Output lands as a reviewable template, never straight into a customer-facing document

CRM sync

• Native bi-directional sync for Salesforce and HubSpot
• Watermark-based incremental delta pulls
• Idempotent upserts with conflict resolution
• Audit logging on every write, encrypted per-tenant credentials

NLU quote assistant

• Reads plain-language input and extracts intent: products, quantities, terms
• Pricing stays in deterministic code, never the model
• Prompt-injection defenses in front of every model call
• Clarification flow when input is ambiguous, backed by a state machine

Go-to-market surface

• Next.js marketing site at campusdyno.io
• AEO content engine publishing reference articles built to be cited by answer engines

Observability

Datadog APM with custom business metrics, Sentry, and structured logging with trace correlation.