Product Requirements Document (PRD)

Solaris ROI Tracking & Value Engine (Lite Edition)

Document Metadata Details Project Code SOL-LITE-ROI Version 1.0 (MVP) Status Approved for Development Context Small-Scale Consulting / Low-Volume SaaS (Max 30 Tenants) Primary Goal Operationalize AI value by converting employee time-savings into verified financial metrics without exposing salary data.

1. Executive Summary

1.1 Problem Statement

Enterprises investing in AI transformation lack credible data to justify the spend. "Soft metrics" (e.g., tokens generated) do not satisfy CFOs. Furthermore, accurate financial calculation typically requires exposing sensitive salary data, which prevents the deployment of broad ROI tools to general staff.

1.2 Proposed Solution

A serverless, multi-tenant ROI Tracking System designed for boutique consulting agencies. It utilizes a "Black Box" Rate Injection architecture where employees log time saved via a gamified wizard, and the server applies hidden financial rates on the backend. This ensures privacy while delivering audit-ready financial reports.

1.3 Scope & Constraints

Target Capacity: Optimized for ~30 active tenants (approx. 500 users/tenant). Infrastructure: 100% Serverless (Scale-to-Zero cost model). Privacy: Hourly rates must never be exposed to the client-side browser. 2. User Personas

Persona Role Motivation Key Pain Point The Maker (Alex, Marketing) Employee Logger Wants recognition for efficiency; hates administrative friction. "If logging takes more than 1 minute, I won't do it." The Checker (Sarah, Dept Head) Manager Verifier Wants accurate team metrics; fears data inflation. "I don't have time to audit every single claim. I only care about the big ones." The Viewer (David, CFO) Dashboard Consumer Wants defensible numbers to justify budget; ignores "fluff." "Show me the net savings. If the data is self-reported, how do I know it's real?" 3. Functional Requirements

3.1 Epic: The Impact Logger (TurboTax-Style Wizard)

Goal: Reduce the friction of data entry to under 30 seconds per log.

ID Priority Feature Description Acceptance Criteria LOG-01 P0 Wizard UI A multi-step modal asking one question at a time (What -> How Much -> How Often). Must complete in <4 clicks. Progress bar visible. LOG-02 P0 Time Normalization Input accepts minutes or hours; converts to standardized 'minutes' integer for backend. User inputs "1.5 hours", system stores 90. LOG-03 P1 Live Feedback As user drags time sliders, show a "Time Saved/Year" counter updating in real-time. Counter updates locally without API calls. LOG-04 P1 Tool Attribution Dropdown to select the AI tool used (e.g., ChatGPT, Claude, Copilot). List is configurable per tenant. 3.2 Epic: The Black Box Engine (Calculation & Privacy)

Goal: Calculate financial value without exposing salary rates to the frontend.

ID Priority Feature Description Acceptance Criteria CALC-01 P0 Rate Injection Edge Function retrieves secure rate from department_rates table based on user's Dept ID. Rate is NEVER sent in JSON response. Calculation happens 100% server-side. CALC-02 P0 Annualization Formula: (Minutes/60) * Frequency_Multiplier * Rate. Weekly = x52, Daily = x260, Monthly = x12. CALC-03 P2 Snapshotting Store the calculated hourly_rate used at the moment of creation in the log row (hidden) to preserve historical accuracy if rates change later. Logs retain value even if Dept Rate is updated next month. 3.3 Epic: Governance & Verification

Goal: Ensure data credibility through a "Maker-Checker" workflow.

ID Priority Feature Description Acceptance Criteria GOV-01 P0 High-Value Trigger If annualized_value > $5,000 (configurable), set status to pending_verification. Low value logs auto-verify. High value logs lock until actioned. GOV-02 P1 Manager Queue View showing pending logs for the manager's specific department only. Manager can 'Approve' or 'Reject' with comment. GOV-03 P2 Email Alerts Async notification via Resend when a high-value log is submitted. Email contains "Approve" deep link. 3.4 Epic: The Value Dashboard

Goal: Visualize ROI for the consulting client/CFO.

ID Priority Feature Description Acceptance Criteria VIS-01 P0 Aggregates Display Total Annualized Savings ($) and Hours Repurposed. Filters: Date Range, Dept, Tool. VIS-02 P1 ROI Multiplier Dynamic card showing (Total Savings / Consulting_Retainer_Cost). Shows "5x ROI" etc. VIS-03 P2 Leaderboard "Top Savers" list to gamify adoption. Anonymized option available in settings. 4. UX & Workflows

4.1 "TurboTax-Style" Logger Flow

Design Philosophy: Cognitive ease. One question per screen.

Screen 1: The Win Input: Workflow Name (Text) + Tool Used (Icon Grid). Action: "Next" Screen 2: The Impact Input: Time Saved (Slider: 5min to 4hrs) + Frequency (Toggle: Daily/Weekly). Visual: "That's 120 hours saved per year!" (Calculated locally). Action: "Next" Screen 3: Summary Display summary card. Action: "Submit Win". System Response: Confetti animation. "Value Secured: $pending" (Do not show specific $ if rate is hidden, or show aggregate only). 4.2 Verification Logic Gate

Value < $5k Value >= $5k Approve Reject User Submits Log Calculate Value Status: Verified Status: Pending Dashboard Update Email Manager Manager Action Status: Rejected Notify User 5. Technical Architecture & Data Model

5.1 Tech Stack (Low-Cost / Serverless)

Frontend: Next.js (hosted on Vercel). Backend: Supabase (PostgreSQL + Auth + Edge Functions). Visualization: Tremor (React components). Email: Resend. 5.2 Schema Specifications (PostgreSQL)

Table: department_rates (RLS: Protected)

Crucial: This table must have RLS enabled so anon and authenticated roles cannot SELECT without specific admin filters. TABLE department_rates ( id uuid PRIMARY KEY, tenant_id uuid, dept_name text, -- 'Finance', 'Engineering' hourly_rate numeric -- SENSITIVE: $65.00 ); Table: impact_logs

TABLE impact_logs ( id uuid PRIMARY KEY, tenant_id uuid, user_id uuid, minutes_saved int, frequency text, annualized_value_money numeric, -- Populated by Server-Side Function only status text DEFAULT 'verified' -- 'pending_verification', 'verified', 'rejected' ); 5.3 Security Implementation

Row Level Security (RLS): Employee: Can INSERT logs. Can SELECT only own logs. Cannot SELECT rates. Manager: Can SELECT logs where user.dept_id == manager.dept_id. Admin: Can SELECT/UPDATE all. Edge Function Logic: Calculation logic resides in a Supabase Edge Function (Deno/TypeScript), not in the Next.js API routes, to ensure strict environment variable isolation for DB service keys if needed (though RLS is preferred). 6. Non-Functional Requirements

6.1 Performance

Latency: Dashboard initial load < 1.5s (via Supabase query optimization). Write Speed: Log submission < 500ms. 6.2 Scalability & Cost

Scale-to-Zero: System costs must be near $0 when idle (nights/weekends). Concurrency: Must support 30 concurrent writes (1 per tenant) without locking. 6.3 Privacy & Compliance

Data Isolation: Strict tenant_id enforcement on EVERY query. Currency: Single currency per tenant (defined at onboarding) to reduce complexity. 7. Success Metrics & KPIs

Metric Definition Success Threshold Adoption Rate % of licensed users submitting at least 1 log/month. > 40% Verification Velocity Avg time for Manager to verify a pending log. < 48 Hours Value Realization Total Verified Savings vs. Platform Cost. > 5x ROI 8. Appendix: Implementation Plan

Phase 1: Foundation (Week 1)

Setup Supabase Project & Auth. Implement Tables & RLS Policies (Critical Path). Manual seed of department_rates. Phase 2: The Engine (Week 2)

Develop calculate-impact Edge Function. Implement Unit Tests for calculation logic (ensure math is perfect). Phase 3: Experience (Week 3)

Build Next.js "TurboTax" Wizard. Build Manager Verification Queue. Phase 4: Visualization (Week 4)

Build Tremor Dashboard. End-to-End User Testing. Deploy to Vercel Production.