Customer scope builder

Tell us what the system can do.

Select one primary architecture and the capabilities layered on top. If you do not know the architecture, begin with a safe voice discovery profile and refine it before production testing.

One primary patternComposable add-onsVisible price logicNo production writes by default

Do you know the architecture?

Choose the closest pattern. It can be corrected during discovery.

Architecture knowledge

Safe discovery default

Cascaded voice assessment

We begin with synthetic inbound calls, provider-neutral speech stages, no write-capable tools, no retained customer PII, a human handoff path, safety tests, and complete tracing.

Twilio-compatible telephony contract
STT → LLM → TTS stage isolation
Latency, silence, interruption, and handoff tests
Discovery findings before production access

Add the capabilities it uses.

Each selection expands the risks, evidence, and estimate.

RAG knowledgeRetrieval, freshness rules, reranking, and citations.OptionalAction-taking toolsTyped tool calls that read or change business state.OptionalLong-term memoryCustomer or organizational memory retained across sessions.OptionalHuman handoffEscalation, approval, or transfer to a person or queue.IncludedModel/provider fallbackRetries, alternate providers, circuit breakers, and degraded modes.OptionalSafety guardrailsInput/output controls, policy checks, and protected-topic handling.IncludedTrace observabilityModel, retrieval, tool, cost, and latency traces such as Langfuse.IncludedPII and complianceConsent, disclosure, redaction, retention, and regulated-data boundaries.Optional

Preliminary estimateDiscovery profile

$15,000

Advanced agent audit

Start with $1,500 audit readiness if the customer is not ready for the full voice engagement. It is credited toward an approved audit.

Base engagement$15,000

Price compositionBase vs. selected add-ons

Base package Add-ons

Generated audit scopeUnique requirements

Test packs19

Evidence fields20

Add-ons3

Before this becomes a quote

We confirm access, case volume, sensitive data, languages, environments, and remediation responsibility. Vendor usage, taxes, and unusual compliance work are separate.

Review required information See benchmark evidence

Technical manifest

{
  "schemaVersion": "1.0",
  "discoveryMode": true,
  "primaryArchitecture": "cascaded-voice",
  "addons": [
    "handoff",
    "guardrails",
    "observability"
  ],
  "requiredTestPacks": [
    "Webhook contract",
    "Media format",
    "STT entities",
    "Latency",
    "Barge-in",
    "TTS",
    "Handoff",
    "Escalation triggers",
    "Summary fidelity",
    "Routing",
    "Approval enforcement",
    "Prompt injection",
    "Data exfiltration",
    "Policy bypass",
    "Safe refusal",
    "Trace completeness",
    "Correlation",
    "Cost accounting",
    "Latency budget"
  ],
  "requiredEvidence": [
    "Call SID",
    "Stream events",
    "Audio fixtures",
    "Transcript",
    "Turn timings",
    "Tool calls",
    "End reason",
    "Handoff reason",
    "Summary",
    "Destination",
    "Human outcome",
    "Policy version",
    "Guardrail decision",
    "Blocked content",
    "Response",
    "Trace ID",
    "Span tree",
    "Token usage",
    "Cost",
    "Timing"
  ],
  "preliminaryEstimateUsd": 15000
}