How to build an AI CRM/sales agent
An end-to-end walkthrough of designing, deploying, and hardening an AI-native CRM/sales agent: from ingesting customer data and wiring tool access to evaluating dialogues, enforcing policies, and measuring impact in production.
Nov 20, 2025 · Online
Remote session
Presentation
Recording
Open on YouTube →Talk overview
- Architecture for syncing CRM data, summarizing accounts, and exposing safe tool adapters for email, calendar, and deal updates.
- Conversation flows that mix scripted intent detection with multi-turn reasoning, guardrails, and retrieval for compliance-grade responses.
- Evaluation strategies (simulated prospects, red-team prompts, latency/cost dashboards) to ship reliably and iterate post-launch.