Sales Call Intelligence & Coaching Agent
Multi-agent systems that prospect, personalize, listen, and forecast. These are the hardest problems in the booklet — they demand reasoning under uncertainty, real business-tool integration, and agents that must earn trust before a human hands them outreach or revenue decisions.
Background
Sales managers cannot review every call their team makes, so valuable coaching moments, competitor mentions, and early deal-risk signals get missed. Reviewing calls manually also doesn't scale as a team grows.
Objective
Build an agent that processes sales call transcripts, extracts key moments (objections, pricing discussion, competitor mentions, next steps), scores overall call quality, and generates specific, actionable coaching feedback for the rep — plus a deal-risk flag where relevant.
Key Features
1. Transcript ingestion (from provided text transcripts, or audio via speech-to-text) 2. Moment-extraction agent identifying objections, pricing talk, competitor mentions, and next-step commitments 3. Talk-ratio and filler-word/pace analysis 4. Coaching-feedback agent generating specific, timestamped, actionable feedback tied to real moments in the call 5. Deal-risk flag based on signals like unclear budget/authority/need/timeline (BANT-style gaps) or hesitation language
Expected Solution
A parsing agent that structures the raw transcript into moments and speaker turns An analysis agent layer for both quantitative (talk ratio) and qualitative (sentiment, objections) signals A coaching-report generator that ties every piece of feedback to a specific transcript moment, not a vague generality
Suggested Technology Stack
Speech-to-text (if audio): Whisper (open-source) Extraction/classification: LLM few-shot extraction (no fine-tuning required) Orchestration: LangGraph Analysis: Pandas for talk-ratio/pace metrics Frontend: Streamlit / Plotly dashboard per call and per rep
Expected Outcomes
Consistent, scalable coaching feedback for every call, not just the few a manager has time to review Earlier detection of deals at risk due to unresolved objections or missing stakeholders A searchable library of coaching insights across the whole sales team
Possible Use Cases
A SaaS sales team scaling coaching without scaling management headcount A call center running consistent QA across hundreds of calls a week A sales enablement lead building a training library from real call moments
Evaluation Parameters
Accuracy of moment extraction (objections, competitor mentions, commitments) Specificity and actionability of the coaching feedback (not generic advice) Correctness and usefulness of deal-risk flags Dashboard clarity for both reps and managers