Sintegrum: AI Assessment of Sales Calls and Rankings

Project context

Integrated AI into a multi-tenant CRM to assess how sales teams communicate with customers: whether they ask appropriate questions and correctly interpret the answers. Assessment results fed branch ranking tables.

My engineering contribution

• Designed asynchronous call processing, prompt orchestration, structured AI responses and reporting APIs.
• Implemented RabbitMQ jobs with idempotent processing and reliable error handling.
• Designed migrations and indexing strategies for isolated branch databases; integrated AI, telephony, messaging and accounting services.

Results & scope

Delivered a workflow from AI integration and asynchronous processing to structured sales communication assessments and branch ranking tables. The rankings supported comparison of sales teams by communication quality; no revenue uplift is claimed. Participated in architecture decisions, API design and code reviews.

This is my part of the team's work, measured in that project. Your baseline will be different, and I measure it before changing anything.

Technologies used

• PHP 8
• Laravel
• Yii2
• Node.js
• MySQL
• Redis
• RabbitMQ
• JWT
• OpenAPI
• Docker
• Codeception

Practical reading

AI for calls in a CRM: start with one decision

Multi-tenant data isolation: where the leak starts