AI Sales Analytics Assistant: Reliable Answers in Seconds
reporting sources unified in one assistant
of answers grounded in live sales data
sales dimensions compared in a single question
About the Client
The company (NDA) is a large international manufacturer supplying showrooms and designers worldwide.
The client and BotsCrew have partnered since 2023, beginning with an internal AI project that has become part of employees' everyday work. As the partnership grew, BotsCrew developed a deeper understanding of how the client's teams work day-to-day and identified a new opportunity on the sales side of the business: helping leaders move faster from data to decision.
The Challenge
The client's sales leaders had access to performance data, but every business question took too much manual work to answer.
Leadership needed to compare performance across regions, reps, brands, product categories, showrooms, and time periods.
Answers were spread across long PDF reports (some reaching 100+ pages), spreadsheets, an internal reporting portal, BI dashboards, and CRM data.
One answer often meant switching between systems, searching long reports, comparing figures by hand, or waiting for someone else to pull the data.
The workflow supported static reporting, but it wasn't designed for fast, on-demand sales analysis.
Our Approach
We started with a Discovery Workshop to map the full picture before building anything: the pain points behind leadership's questions, the real use cases the assistant had to cover, and the as-is state of the client's reporting workflow.
The discovery confirmed that the key challenge was preparing the client's sales database for reliable AI-driven analysis. The database was complex: multiple connected tables, business-specific definitions, regional codes, and sales figures that change as new orders are processed.
To make the assistant reliable, BotsCrew focused on three layers:
- Mapping the data structure. We mapped how tables, fields, and relationships connected across the sales database.
- Translating data into business language. We clarified what key fields, acronyms, and sales terms meant in the client's own business context.
- Building the training layer. We built that knowledge into the assistant so it could turn real sales questions into reliable database queries.
How the Solution Works
BotsCrew built a secure AI Sales Analytics Assistant that turns plain-English sales questions into database-backed answers.
- Ask a sales question. A sales or regional leader asks a question in plain English, such as: "What are sales in the Northeast region in 2025 by product category?"
- Understand the business context. The assistant uses company-specific table descriptions, column definitions, relationships, and acronyms to understand the question.
- Generate the SQL query. The AI creates the query needed to retrieve the right answer.
- Run it against live sales data. BigQuery runs the query against the client's live sales database.
- Return a clear result. The assistant returns a direct answer with a comparison table or basic chart.
Sensitive sales and financial data stay inside the client's environment and are never sent to an external LLM.
Key Features
Sales leaders ask about regions, reps, brands, products, showrooms, or time periods, with no SQL or BI filters.
The assistant compares across dimensions to show top and underperforming products, regions, and reps.
Every answer comes from a query on the client’s own sales database, with tables and charts ready for quick reporting.
Analysts manage datasets, define terms and acronyms, train the assistant on real questions, and review feedback, with no engineering support.
Each user sees only the data their role allows.
Power users can view the SQL behind any answer to validate results and improve future responses.
Under the Hood
- BigQuery
- Live sales database
- Gemini
- Text-to-SQL
- Training layer
- Database-only architecture
- Role-based access control
- Data stays in client environment
- Admin console
- Debug mode with SQL view
- Feedback review
What Made It Work
- Built for in-house ownership. The client's analysts can update dataset descriptions, acronyms, and trained questions, and act on user feedback themselves, without relying on developers for every change.
- Easier to maintain as the business changes. As regions, reps, products, or reporting needs evolve, the assistant is updated through the control layer instead of becoming outdated like a static report.
- Answers leadership can trust. With transparent query logic and analyst-managed training, the team has a clear way to understand, validate, and improve the assistant's answers.
- A base for deeper sales intelligence. The assistant gives the client a secure foundation for future analytics use cases as sales reporting needs become more advanced.
Key Results
For the client, the AI Sales Analytics Assistant became the next step in a long-term AI partnership with BotsCrew: from internal AI tools for employees to secure, data-grounded analytics for sales leadership.
What's Next?
Future direction discussed with the client (not yet implemented):
- Proactive insights: automated alerts when a product line declines for several periods in a row
- Industry benchmarking: comparing the client's performance against industry averages
- Continued improvements in performance and response time
Partner with BotsCrew to build AI that is fair, transparent, and aligned with your company's values.
Contact us
