AI Sales Coach

The ElevenLabs GTM Enablement team built an AI role-play coach based on ElevenAgents to solve the training problem for its rapidly growing global sales organization. Existing slides, recorded demos, and peer role-play had inconsistent quality, were hard to measure, and were difficult to coordinate across regional time zones.

The coach consists of voice-based simulations that behave like real enterprise buyers. Each persona hides information across multiple layers, and only when a rep asks the right questions does deeper information surface, such as compliance requirements, past vendor failures, or budget constraints.

Evaluation and feedback are also structured. Conversations follow a fixed beat structure, and once a session ends, the agent drops the character and scores technical accuracy, buyer credibility, and conversation altitude on a 1-3 scale. Structured data is automatically extracted from every transcript and used for leaderboards and team-level skill gap analysis.

If a rep gets stuck midway, they can switch to teaching mode at any time for help.

• The coach pauses the scenario
• briefly explains the needed skill
• provides example phrasing
• and then resumes the conversation

The results were clear. The program was rolled out globally with hundreds of reps participating, and completion rates rose from 40-60% for peer role-play to over 90%. Participants gave it a 5/5 satisfaction rating, and enablement leaders could, for the first time, use aggregated beat-level data to identify which conversation stages were weak.

The next step is deeper performance measurement. The team plans to use beat-level evaluation and Data Collection to map key competency gaps, and to track how program graduates differ from non-graduates in discovery-to-trial conversion, win rate, and deal size.

Ultimately, the plan is to move beyond generic buyer personas toward deal-specific practice. By drawing on LinkedIn, company history, industry, and CRM data, the coach will impersonate the actual person a rep is about to meet, offering pre-meeting rehearsal and post-meeting coaching that revisits missed opportunities-turning AI role-play into a continuous learning layer wrapped around the entire pipeline.