Demo Agent Customization

The demo agent adapts to your product, your buyers, and your go-to-market motion. Out of the box it runs a solid demo, but customization is what makes it feel like your SE rather than a generic AI.

All customization happens in the Demo Configuration section of your site in the dashboard. Changes apply to all new demo sessions immediately.

Voice personality

The agent's voice personality controls how it sounds during conversations. This isn't about the literal voice (pitch, speed), but the communication style the agent uses when explaining features, answering questions, and guiding the prospect.

Personality presets

Choose a starting point:

• Warm: Friendly, conversational, uses casual language. Good for products sold to individual contributors or small teams. The agent says things like "Let me show you something cool" and "This is where it gets interesting."
• Professional: Balanced, clear, business-appropriate. Good for mid-market and enterprise sales. The agent keeps explanations crisp and structured.
• Technical: Detail-oriented, precise, comfortable with jargon. Good for developer tools, infrastructure, and technical buyers. The agent goes deeper on how things work under the hood.

Custom personality instructions

Beyond presets, you can write free-form instructions that shape the agent's behavior:

• "Always explain features in terms of time saved"
• "Use the prospect's company name naturally in conversation"
• "When discussing integrations, mention that we have a public API"
• "Keep explanations under 30 seconds before pausing for questions"

These instructions layer on top of the personality preset. The agent follows them across the entire session.

Product emphasis

Not all features deserve equal airtime. Product emphasis controls where the agent spends its focus during demos.

Priority areas

In the configuration, you can mark product areas as:

• Lead with: The agent opens with these. They're your strongest selling points. If the prospect doesn't have specific requests, this is what they see first.
• Show when relevant: The agent brings these up when the conversation naturally goes there. The prospect asks about reporting, and the agent shows reporting.
• Mention briefly: The agent acknowledges these features exist but doesn't demo them in depth. "We also have audit logging, which I can show you separately."
• Skip entirely: Features the agent never brings up or navigates to. Use this for beta features, deprecated sections, or areas that confuse prospects.

Use case framing

You can configure how the agent frames your product for different buyer types:

• "For HR buyers, lead with employee engagement surveys and skip the developer API section"
• "For IT buyers, emphasize SSO, audit logs, and compliance features"
• "For founders, focus on time-to-value and self-serve setup"

The agent uses prospect context (role, company, stated interests) to select the right framing. If no context is available, it defaults to the general priority ordering.

Guardrails

Guardrails define topics the agent should avoid or handle carefully. These protect your sales process from the agent saying something it shouldn't.

Topic restrictions

Configure topics the agent will not discuss in detail:

• Pricing: The agent can acknowledge that pricing exists ("We have flexible pricing based on usage") but won't quote specific numbers or negotiate. Instead, it steers toward a sales conversation.
• Competitor comparisons: The agent won't compare your product to named competitors. If a prospect asks "How are you different from [Competitor]?", the agent focuses on your strengths without directly referencing the competitor.
• Internal roadmap: The agent won't discuss unreleased features or timelines. If asked about upcoming features, it can say "We're always improving the platform" and redirect.
• Custom topics: Add any other restricted areas. "Don't discuss our recent security incident," "Don't mention the legacy API," "Don't speculate about enterprise pricing tiers."

Handling restricted topics

When the agent encounters a restricted topic, it doesn't refuse awkwardly. It acknowledges the question and redirects:

"That's a great question about pricing. I want to make sure you get the right answer based on your specific needs. Let me connect you with someone on the team who can walk through options with you."

This feels natural rather than evasive. The goal is to keep the demo flowing while protecting sensitive areas.

Accuracy guardrails

Beyond topic restrictions, you can set accuracy rules:

• "Never claim we have SOC 2 certification" (if you don't)
• "Don't say the product supports more than 10,000 users per account"
• "Always clarify that the mobile app is in beta"

These prevent the agent from making claims your product can't back up.

Discovery questions

Discovery questions are what the agent asks early in the session to understand the prospect's needs, similar to a discovery call. Good discovery makes the rest of the demo feel personalized.

Default discovery flow

The agent asks 2-3 discovery questions at the start of each session:

1. Role question: "What's your role at [Company]?" Helps the agent tailor feature emphasis.
2. Pain point question: "What's the biggest challenge you're trying to solve?" Drives which capabilities the agent prioritizes.
3. Context question: "Are you evaluating other solutions, or is this your first look at this space?" Helps the agent calibrate how much background context to provide.

Customizing discovery

You can replace or add to the default questions:

• "Ask what tools they currently use for onboarding"
• "Ask how many users they need to support"
• "Ask whether they've tried solving this problem before and what happened"

Keep discovery to 2-4 questions. More than that feels like an interrogation, and the prospect came to see the product.

Using discovery data

The agent uses discovery answers throughout the session. If the prospect says they're struggling with user activation, the agent emphasizes features related to activation and time to value. If they mention they have a large team, the agent highlights admin controls and permissions.

Discovery isn't just polite conversation. It shapes the entire demo.

Combining customizations

Customizations work together. A technical personality with emphasis on integrations, guardrails against pricing discussion, and discovery questions about the prospect's tech stack produces a very different demo than a warm personality emphasizing ease of use with broad feature coverage.

Test different combinations by running sessions from the dashboard. The goal is a demo that feels like your best SE on their best day.

What's next

• Demo agent overview for session intelligence and post-demo data
• Demo agent setup if you haven't configured the basics yet

FAQ

Can I have different configurations for different buyer personas? Not as separate saved profiles yet. You can change settings between sessions, and the agent adapts based on prospect context you provide when starting a session.

How specific can guardrails be? Very. You can set broad topic restrictions ("no pricing") or narrow rules ("don't mention the legacy REST API, only mention the GraphQL API"). The agent follows both.

What if the prospect asks a discovery question back? The agent handles reverse questions naturally. If the prospect asks "What do most companies your size use this for?", the agent answers based on your content before continuing with its own discovery.

Can I see how customizations affect demo quality? Each session captures a full transcript and engagement signals. Compare sessions with different configurations to see which produces better conversations and more buying signals.