SQL tables - creation of SEO dashboards and AI analytics

Short description of the task: there are SQL tables (hereinafter DB), it is necessary to connect AI to them. As a result:

1. Its (AI) opinion on the data as a whole, its opinion on the data regarding "negatives", its recommendations, and its opinion on specific point queries, for example, "why is product 123 selling poorly?", "what actions are recommended to increase sales of product 123?".

2. Building a dashboard (Google Sheets or another system of your choice) with specific indicators (about 100 indicators) on a periodic basis (manual launch or according to a schedule)

Important: the results of the AI's work must be free of hallucinations - the correctness check is verified on a reduced number of rows in the DB.

Detailed description of the task:

1. Main tasks:

- CEO Dashboards: Creating high-level reports for management. Key metrics: P&L, Cash Flow, sales analytics (LFL, plan-fact), inventory analysis, and marketing effectiveness.

- AI Data Agent: Development of an AI bot (based on LLM) that understands the structure of our DB and provides analytical responses to queries in natural language (for example: "What is the margin for brand X for the past week across channels?")

- Data Integration: Setting up automatic data collection and cleansing from SQL tables. Preferably from multiple sources.

2. Technological stack and sources: SQL (MS SQL Server): Working with a large database (>50 tables, millions of records).

3. Specifics of our database:

- Our DB includes over 60 tables from several information collection sources - CRM, website, BAS, etc.

- Key data blocks:

Sales and orders: separate tables for sales

Finance: tables with cash movements and balances

Clients and loyalty program: tables with clients and bonus movements

Warehouses and nomenclature: tables with stock balances and product movements

4. Candidate requirements:

- Expert SQL: writing queries to the DB through AI. In perspective, optimization of queries to the DB by reducing the number of tokens used

- Skills in working with AI: experience in creating AI agents that can work with databases (RAG, Text-to-SQL).

- Understanding of business metrics: metrics EBITDA, ROMI, Churn rate, ARPU

5. The AI agent should not just return dry numbers from SQL, but act as an Executive Assistant, which includes:

- forming "Own Opinion" (Data-Driven Insights):

- the ability to interpret results. For example: "Sales in category X increased by 10%, but profit fell due to rising manual cost in the table items_manual_cost_of_goods_sold. This indicates inefficiency in procurement".

- critical analysis (Identification of Risks/Downsides):

- AI must independently find anomalies and "weak spots". For example: "The downside of the current strategy is a high return rate compared to the previous period, which negates the benefit from sales volume".

- proposals based on correlations. For example: "Based on the data from Site_promo_code and sales, I recommend stopping promotion X, as it attracts customers with low LTV, and reallocating the budget to the loyalty program (clients_pl)".

We work with individual entrepreneurs