To understand what drives success, Hightouch employs LLMs to synthesize all interaction data for an account from Gong, Salesforce, and Slack. The model then generates a qualitative narrative, or "deal story," explaining the deal's progression, providing deeper insights than traditional attribution.

Leaders often believe their data is adequate until they attempt to deploy an AI agent. The process quickly reveals years of inconsistent or missing data from sales teams, forcing a critical data hygiene cleanup that should have happened long ago.

The most advanced GTM teams are abandoning traditional CRMs like Salesforce as their primary interface. Instead, they use data warehouses (Snowflake, Databricks) for flexible data storage and push curated insights to reps directly within their workflows (Slack, email, Notion), eliminating the need for manual data entry and retrieval.

Despite having a large staff, Gary found crucial context was lost from meetings. He now uses an AI tool as a "capture all" CRM, sending it photos and notes via text. The AI builds a relationship graph that he then uses to automate follow-ups and maintain connections, essentially scaling his personal memory.

Forward-thinking companies follow a "data-first" strategy, ingesting intent data into a central data lake (e.g., Snowflake) alongside CRM and call data. This creates a unified source of truth that can be queried by AI agents (e.g., Claude), empowering account executives to ask complex, contextual questions and get immediate answers.

Structure your CRM to minimize clicks and context switching for SDRs. Create a single, clean view showing a list of accounts with all relevant contacts and their data on one screen. This turns the CRM from a passive database into an active, high-efficiency prospecting workspace.

Christopher O'Donnell's new company, Day AI, is building a CRM from the ground up to be "LLM optimized." Unlike traditional CRMs that resemble spreadsheets, it ingests and stores all company interactions in a way that allows an AI agent to easily explore the network of relationships and answer complex, natural language questions instantly.

Pre-call research that used to take 20 minutes can now be automated. Use the 'deep research' function in AI tools, connecting them to your CRM and the public web to get a detailed brief on a company and contact. Always have a human review the final output.

LLMs dramatically accelerate market research but are non-deterministic and lack real-world grounding. Their true value is preparing for customer conversations—crafting questions, understanding market history, and practicing listening. They augment human judgment, they don't replace it.