Spendora

AI tooling cost optimization for modern teams.

Spendora analyzes AI subscription stacks, detects operational inefficiencies, estimates potential savings, and generates shareable audit reports for teams using tools like ChatGPT, Claude, Gemini, Cursor, Copilot, and Windsurf.

Demo

Live App: https://spendora-aibuddy.vercel.app/

Screenshots

Landing Page

Audit Report

Audit Dashboard

The Problem

AI tooling adoption inside startups and engineering teams has accelerated rapidly. Teams often subscribe to multiple overlapping products such as ChatGPT, Claude, Gemini, Cursor, GitHub Copilot, and Windsurf without clear operational visibility into usage efficiency or subscription overlap.

As AI tooling stacks grow:

• redundant subscriptions increase
• seat allocation becomes inefficient
• pricing structures become fragmented
• optimization opportunities become difficult to identify

Most teams currently manage AI spend manually through spreadsheets or finance reviews, which creates operational blind spots and unnecessary monthly costs.

The Solution

Spendora provides a deterministic AI tooling audit system that:

• analyzes subscription configurations
• estimates optimization opportunities
• identifies overlapping tooling patterns
• generates executive summaries
• creates shareable audit reports
• enables authenticated audit history and persistence

The platform combines:

• deterministic pricing analysis
• operational heuristics
• AI-generated executive summaries
• persistent audit ownership

Core Features

AI Spend Audits

Analyze AI tooling stacks across products such as:

• ChatGPT
• Claude
• Gemini
• Cursor
• GitHub Copilot
• Windsurf

Deterministic Pricing Engine

Spendora uses centralized pricing logic and threshold-based analysis to estimate:

• monthly savings
• annual savings
• operational inefficiencies

AI Executive Summaries

Audit reports include AI-generated executive summaries with deterministic fallbacks for reliability.

Authenticated Audit Ownership

Users can:

• create accounts
• save audits
• revisit historical reports
• manage previous optimization workflows

Shareable Reports

Every audit generates:

• public URLs
• OG metadata
• social sharing previews

Audit Dashboard

Authenticated users can:

• view historical audits
• reopen reports
• manage optimization history

Lead Capture Workflow

Audit reports support:

• email sharing
• lead capture
• team collaboration workflows

Architecture Overview

Spendora is built as a full-stack App Router application using:

• Next.js 16
• TypeScript
• Supabase
• OpenRouter
• TailwindCSS
• Jest
• Zod runtime validation

The architecture emphasizes:

• deterministic business logic
• thin API orchestration
• centralized pricing systems
• typed persistence
• authenticated ownership
• production-safe validation

For a deeper system walkthrough, see ARCHITECTURE.md.

Request Lifecycle

User submits audit
→ request validation
→ deterministic audit engine
→ AI summary generation
→ Supabase persistence
→ shareable report rendering
→ authenticated ownership (optional)

---

Tech Stack

Frontend

• Next.js 16 App Router
• React
• TypeScript
• TailwindCSS

Backend

• Route Handlers
• Supabase
• Zod validation
• Deterministic audit engine

AI

• OpenRouter
• AI-generated summaries
• fallback summary pipeline

Testing

• Jest
• Runtime validation tests
• API route tests
• Audit engine tests

---

Local Development

Install Dependencies

npm install

Start Development Server

npm run dev

Run Quality Checks

npm run lint
npm run type-check
npm test
npm run build

---

Environment Variables

Create a .env.local file:

NEXT_PUBLIC_SUPABASE_URL=
NEXT_PUBLIC_SUPABASE_ANON_KEY=
SUPABASE_SERVICE_ROLE_KEY=

OPENROUTER_API_KEY=

RESEND_API_KEY=
FROM_EMAIL=

---

Product Decisions

Anonymous-First Onboarding

Users can run audits before authentication to reduce onboarding friction and improve conversion.

Deterministic Audits Over Fully AI-Generated Analysis

Spendora prioritizes deterministic pricing logic for:

• consistency
• operational trust
• lower inference costs
• predictable outputs

AI is used only for executive summaries and enrichment.

Centralized Pricing Architecture

All pricing logic is centralized in pricing.ts to avoid fragmented business rules and simplify maintenance.

Thin Route Handlers

API routes are intentionally minimal and delegate business logic to isolated library modules.

---

Stability & Verification

The project currently includes:

• 58 passing tests
• full TypeScript validation
• ESLint validation
• successful production builds
• authenticated persistence flows
• SSR-compatible App Router architecture

Verified flows:

• audit generation
• AI summary generation
• lead capture
• authenticated dashboards
• audit sharing
• audit ownership linking

---

Future Improvements

Potential future improvements include:

• smarter recommendation reasoning
• recurring audits
• CSV exports
• analytics dashboards
• team workspaces
• richer overlap detection
• recommendation confidence tuning

---

Reflection

The most difficult parts of the project were not UI implementation, but:

• authenticated ownership flows
• SSR-safe auth handling
• deterministic audit modeling
• runtime validation
• anonymous-to-authenticated state continuity

One major product insight from building Spendora was that recommendation trust and operational clarity matter significantly more than AI-generated complexity.

---

License

MIT