Sales Signals app (Agentic AI)

Automated sales coaching engine that turns B2B call transcripts into real-time, context-aware feedback, combining LLMs and historical customer data to surface revenue and retention insights.
App
Agentic AI
Python
Featured
Author

Aleksei Prishchepo

Published

February 5, 2026

Project Overview

This system analyzes sales conversations using LLMs enriched with long-term customer context via a specialized RAG pipeline. By synthesizing current calls with historical interactions, it delivers continuity-aware coaching that reflects relationship momentum rather than isolated incidents. The result is consistent, data-driven revenue enablement grounded in real-world commercial outcomes.

NoteRole

End-to-end owner: system architecture, LLM orchestration, RAG-driven context injection, and interactive analytics dashboard.

NoteTools

FastAPI • Python • LLM APIs • Retrieval-Augmented Generation (RAG) • Altair Viz • Async Event Processing • PostgreSQL/SQLAlchemy

NoteDomain

B2B Sales Analytics • Revenue Enablement • Conversational Analytics • Performance Management

Key Features & Components

Context-Aware Coaching

Generates targeted observations and next-step recommendations, dynamically adjusted based on retrieved historical client context and previous commitments.

Automated Periodic Synthesis

Orchestrates daily and operator-level “roll-up” summaries, distilling high-volume call data into actionable executive insights and individual performance trends.

Advanced Sales KPI Tracking

Calculates sophisticated metrics including Objection Intensity, Discovery Depth, Customer Sentiment, and Explicit Close Attempt Rates to identify behavioral gaps.

Historical Relationship RAG

Heuristic-driven retrieval of caller history ensures the LLM “remembers” previous objections, pricing discussions, and deal momentum.

System architecture diagram

System architecture diagram

System architecture diagram

System architecture diagram

Implementation

Bimodal LLM Orchestration

Lightweight models handle classification and routing, while higher-reasoning models generate nuanced coaching insights.

Asynchronous Pipeline

Event-driven architecture with FastAPI background tasks ensures seamless webhook ingestion from transcription providers.

Interactive Performance Dashboard

Custom Altair-based visualizations allow stakeholders to filter performance by operator, deal stage, and behavioral signals.

Automated Insights Delivery

Markdown-based coaching analyses are automatically rendered into HTML and distributed via email for immediate consumption by sales teams.

Outcomes & Impact

TipNear-Real-Time Feedback

Coaching latency reduced from days to minutes, enabling faster behavioral adjustment.

TipChurn Risk Detection

Early identification of sentiment shifts, objections, and signals tied to retention risk.

TipScalable Sales Expertise

Encodes senior-level B2B sales knowledge into a consistent, automated coaching framework.

Skills Demonstrated

LLM Orchestration • Retrieval-Augmented Generation (RAG) • Context-Aware Synthesis • Sales & Revenue Analytics • B2B Sales Methodology • Asynchronous API Design • Prompt Engineering • Data Visualization (Altair) • Multi-model LLM Architectures • Automated Insight Generation • End-to-End System Design

Apply This to Your Business

Want to scale sales coaching without it becoming generic? Let’s talk about how I help turn real sales conversations into actionable, context-aware insights.

See Also

Back to top