OVERVIEW We are looking for an experienced n8n automation engineer to build an AI-powered daily sales intelligence system for our CEO and executive team. This is an alpha version designed to validate value quickly and cost-effectively — not a full enterprise BI platform. The system will: Ingest daily sales order data from Acumatica (ERP) Ingest daily sales activity + CRM data from HubSpot Store normalized data in a lightweight warehouse Deliver a daily executive summary in Microsoft Teams Allow executives to ask natural language questions in Teams Return deterministic, consistent answers grounded in warehouse data We are Microsoft-based and use Teams + Power BI, but this project focuses on n8n + warehouse + Teams integration. KEY CONSTRAINT Data updates once per day System must be accurate through the previous day No real-time reporting required All outputs must clearly state data timestamp This is intentional to reduce cost and complexity. SCOPE OF WORK DATA INGESTION (V1) Acumatica (ERP) Daily automated emailed reports (CSV/XLSX preferred) Parse and load into warehouse V2: optional API integration (future) Expected fields: Order ID Order date Customer/account Rep Territory SKU / product Quantity Revenue Discounts Order status HubSpot We need: Activity data Calls Emails Text messages (Optional: meetings, tasks) CRM data Companies (accounts) Contacts Deals Owners (reps) Territories (if applicable) Acceptable ingestion approaches: Daily emailed reports Direct API integration We are open to your recommendation based on cost and reliability. DATA WAREHOUSE LAYER (REQUIRED) We need a lightweight warehouse to ensure: Deterministic metrics Consistent answers No hallucinated AI outputs Full audit logging Acceptable V1 options: Azure SQL PostgreSQL Other low-cost solution (justify) Minimum schema concept: fact_orders fact_activities dim_accounts dim_reps dim_territories dim_dates metric_definitions etl_audit_log Requirements: Idempotent loads Row validation Duplicate prevention ERP ↔ HubSpot mapping strategy Full load timestamps AI-POWERED Q&A (GUARDRAILED) Executives will query the system inside Microsoft Teams. Example questions: How many new accounts last week vs week prior? Which territories are down month over month? Show calls by rep yesterday. How many accounts were sold to in the last 30 days? Requirements: AI translates question → structured query Query runs against warehouse Response includes numeric answer, short definition used, and data timestamp No fabricated numbers Clear handling of ambiguous questions We are open to: OpenAI Azure OpenAI Other cost-efficient LLM provider Cost efficiency matters. DAILY EXECUTIVE SUMMARY (PUSH TO TEAMS) Each morning the system should send: Orders + revenue AOV New accounts Recently sold-to accounts Rep activity totals Rep leaderboard Territory summary Exceptions/anomalies Data timestamp + load status MICROSOFT TEAMS INTEGRATION We need: Teams bot or webhook integration Auth-restricted access Query logging Clean formatting Please propose the best architecture. METRIC DEFINITIONS MUST BE CONFIGURABLE Examples: “New account” (HubSpot create date vs first ERP order — definable) “Active account” (X-day activity threshold) “Recently sold-to” (X-day order threshold) Rep attribution logic (HubSpot owner vs ERP rep) Definitions must be stored in a configurable structure (not hardcoded). ALPHA BUILD PRIORITIES Validate executive usefulness Keep infrastructure lean Keep operational costs low Avoid overengineering Ensure reliable daily refresh This is not a full BI ecosystem build. REQUIRED IN YOUR PROPOSAL Recommended Architecture Storage choice n8n workflow structure AI integration approach Teams integration approach Project Plan Phases Estimated timeline Key risks Dependencies Cost Breakdown One-Time Build Cost Design Development Testing Documentation Estimated Monthly Operational Costs (broken out): n8n hosting Database hosting LLM usage (daily summary + moderate exec Q&A) Any additional infrastructure Also include: Lowest-cost viable version Moderate reliability version Scalable future-ready version IDEAL EXPERIENCE Strong n8n experience (complex workflows, error handling) HubSpot API or CRM ingestion experience ERP integration experience (bonus) Data warehouse design experience Experience building LLM-powered structured query systems Microsoft Teams integration experience (strong bonus) WHAT WE’LL PROVIDE Sample ERP daily reports Sample HubSpot reports Access to shared mailbox HubSpot access (if API route chosen) Teams environment details Stakeholder for metric definition approval If responding, please outline your architecture approach, timeline estimate, operational cost estimate, and examples of similar systems you’ve built.