During live sales calls, customers frequently ask detailed product questions that require representatives to manually search internal databases and CRM systems. This process typically takes 25-65 seconds per query, creating awkward pauses that hurt customer experience and reduce sales efficiency. We present SalesCopilot, a real-time AI-powered assistant that eliminates this bottleneck by automatically detecting customer questions, retrieving relevant information from the product database, and displaying concise answers on the representative's dashboard in seconds. The system integrates streaming speech-to-text transcription, large language model (LLM)-based question detection, and retrieval-augmented generation (RAG) over a structured product database into a unified real-time pipeline. We demonstrate SalesCopilot on an insurance sales scenario with 50 products spanning 10 categories (2,490 FAQs, 290 coverage details, and 162 pricing tiers). In our benchmark evaluation, SalesCopilot achieves a measured mean response time of 2.8 seconds with 100% question detection rate, representing a 14xspeedup compared to manual CRM search in an internal study. The system is domain-agnostic and can be adapted to any enterprise sales domain by replacing the product database.
Figures & tables
Figure 1: System architecture of SalesCopilot . The pipeline consists of three layers: the client (React-based dashboard with audio capture), the server pipeline (STT, conversation management, question detection, retrieval, and answer generation), and external services (Deepgram, LLM, SQLite, and ElevenLabs). Components highlighted in amber represent the core AI-driven modules that distinguish SalesCopilot from standard workflows.
Figure 2: The SalesCopilot dashboard during a live sales call. The left panel shows the real-time conversation transcript with color-coded speaker labels (blue for sales rep, green for customer). The right panel displays AI-generated suggestion cards that appear automatically when customer questions are detected, each showing the extracted question, a concise answer with specific product data, the database source, and a confidence score.
Figure 3: Response time comparison between manual search (internal CRM) and measured SalesCopilot latency across six question categories. Error bars indicate standard deviation. Green annotations show the speedup factor. SalesCopilot achieves 9–23 × faster responses depending on question complexity.
Figure 4: Distribution of response times for manual search (left, from internal CRM study) and SalesCopilot (right, measured). Note the different y-axis scales. Manual search exhibits high variance (5–100+ seconds) while SalesCopilot consistently responds within 2–4 seconds.
Figure 5: Measured per-stage latency breakdown by question category. Answer generation (red) is the dominant cost, followed by knowledge retrieval (amber) and question detection (blue). LLM inference accounts for ∼ 71% of total latency.
Figure 6: Cumulative response time as the number of customer questions increases. The shaded green area represents time saved by SalesCopilot compared to manual search (internal CRM study).
Metric
Manual Search
SalesCopilot
Improvement
Avg. Response Time
39.7s (internal study)
2.8s (measured)
14 × faster
Response Time Std. Dev.
12–18s
0.5s
∼ 25 × lower
Question Detection Rate
N/A
100% (20/20)
N/A
Time per 10-Q Call
6.6 min
0.5 min
5.7 min saved
Time per 20 Calls/Day
2.2 hrs
0.2 hrs
1.9 hrs saved
Table 1: Summary of experimental results. SalesCopilot latencies are measured; manual search times are from an internal CRM study.
Figure 7: Overall performance summary. SalesCopilot reduces response time from 39.7s (internal CRM study) to 2.8s (measured), a 14 × speedup, with 100% question detection rate.
Appendix figures & tables4 assets
Supplementary material from the paper’s appendix.
Appendix
Table
Records
Fields
Description
products
50
4
Product name, category, description
coverage_details
290
6
Coverage type, amount, deductible, conditions
policy_terms
50
5
Term length, renewal & cancellation policies
faqs
2,490
4
Question-answer pairs per product
pricing_tiers
162
7
Tier name, monthly/annual premium, age range
Appendix
Table 2: Insurance knowledge base schema and statistics.
Category
Products
FAQs
Coverage
Pricing
Life
5
∼ 250
∼ 30
∼ 16
Health
5
∼ 250
∼ 30
∼ 16
Auto
5
∼ 250
∼ 30
∼ 16
Home
5
∼ 250
∼ 30
∼ 16
Travel
5
∼ 250
∼ 30
∼ 16
Disability
5
∼ 250
∼ 30
∼ 16
Appendix
Table 3: Per-category statistics of the insurance knowledge base.
Category
Product
Example FAQ
Life
SecureLife Premium Term 30
Q: What happens if I get diagnosed with a terminal illness? A: Accelerated benefit up to 50% of death benefit ($250K) if diagnosed with terminal illness.
Health
FamilyCare Advantage Plan
Q: What is the out-of-pocket maximum? A: 6,500/individual,13,000/family. After reaching the limit, the plan covers 100%.
Auto
SafeDrive Elite
Q: Will my premium go up after an accident? A: At-fault accidents increase premiums 20–40%. Accident Forgiveness add-on ($5/mo) waives the first surcharge.
Travel
Globetrotter Shield
Q: Is COVID-19 covered under travel insurance? A: Yes. Medical expenses, trip cancellation due to positive test, and quarantine accommodations are all covered.
Appendix
Table 4: Example products and FAQ entries from the knowledge base.
Type
Direction
Description
transcript_update
Server → Client
Transcript text with speaker label and final/interim flag
suggestion_card
Server → Client
AI-generated answer card with question, answer, source, and confidence
audio_play
Server → Client
Base64-encoded MP3 audio for TTS playback (demo mode)
status
Server → Client
Session status messages (connected, demo started/ended)
error
Server → Client
Error messages
text_input
Client → Server
Manual text input for testing
Appendix
Table 5: WebSocket message protocol between frontend and backend.