Autonomous Sales Agent Boundary Assessment
Analyze commercial negotiation boundaries and pricing guardrails for autonomous customer-facing sales agents.
Deploy this template prior to enabling automated SDR or outbound sales bots in pipeline engagement. It identifies prompt leakage risks, rogue discounting vulnerabilities, and fallback triggers.
Role: VP of Commercial Operations & Autonomous Revenue Systems
Context
- Agent Classification: {{sales_agent_tier}}
- Commercial Limits: {{pricing_authority_matrix}}
- Data Access: {{crm_data_exposure}}
- Market Segment: {{target_market_segment}}
- Competitor Policy: {{competitor_mention_rules}}
- Rep Handoff Triggers: {{escalation_threshold}}
Task
Produce a structured sales agent guardrail analysis that identifies commercial exposure points, prevents unauthorized contractual commitments, and defines automated buyer interaction boundaries.
Method
- Analyze the autonomy level of {{sales_agent_tier}} to expose privilege escalation risks.
- Inspect {{pricing_authority_matrix}} limits against conversational pressure tactics used by buyers.
- Identify unauthorized disclosure pathways for data outlined in {{crm_data_exposure}}.
- Evaluate positioning constraints established in {{competitor_mention_rules}} against common buyer objections.
- Calibrate risk tolerance according to contract stakes within {{target_market_segment}}.
- Formulate conversational deflection scripts and hard validation boundaries for commercial commitments.
- Define dynamic fallback triggers that route interactions based on {{escalation_threshold}}.
Constraints
- MUST NOT allow the agent to quote terms exceeding {{pricing_authority_matrix}} under any phrasing.
- MUST enforce strict prompt-injection defenses against customer-prompted discount overrides.
- Frame all guardrails in the context of commercial liability reduction.
- Maintain high conversion velocity while preventing unauthorized warranty promises.
Output format
- Commercial Vulnerability Breakdown: 3 structured paragraphs covering Pricing, Data, and Positioning.
- Defensive Rule Catalog: Table with 5 rows (Threat Type, Agent Vulnerability, Hard Guardrail, Validation Logic).
- Handoff Workflow: Step-by-step routing protocol for human account executive intervention.
Self-review
- Did I verify that {{pricing_authority_matrix}} contains enforceable upper and lower limits?
- Are data privacy concerns related to {{crm_data_exposure}} resolved with hard filters?
- Is the transition protocol practical for a live sales environment?
Explicit role, a named task, and discrete steps the model can follow.
Background, inputs and variables the model needs before it starts.
Hard boundaries — what the model must and must not do.
A named, field-level shape for the response.
Ordered work items that force analysis before an answer.
Length and structure that travel across frontier models.
Signal density — instruction weight without padding.
Documented variables so the scaffold adapts to new inputs.
Quality bar, assumptions and behaviour when inputs are thin.
How much real usage the template has behind it.