--- title: ORZA-01 — AI CX Voice Agent slug: orza role: AI CX Voice Agent function: Customer / Support seniority: Senior verticals: - "SaaS" - "Consumer" - "Mid-market" tier: expert monthlyPrice: $1000/month publishedAt: "2026-05-04T16:00:00.000Z" canonical: "https://fidelic.ai/agents/orza" --- # ORZA-01 — AI CX Voice Agent *Brand-tuned voice across chat, voice, SMS* ORZA handles brand-voice-tuned conversations across voice, chat, SMS, and in-app channels — with the brand-voice samples your CX lead has approved and a four-tier constitution gating every response. ## Why it matters Sierra runs at $150K+/yr enterprise floor with negotiated outcome-based pricing. ORZA is $1,000/month flat, with a published constitutional gate on every response and brand-voice samples frozen at deployment. ## Capabilities - Handles voice, chat, SMS, and in-app conversations under your brand-voice baseline - Routes by channel and complexity per your CX lead's deployment rules - Escalates with full conversation transcript and the rule that gated the escalation - Surfaces brand-voice drift candidates weekly for your CX lead's review - Logs every response with the brand-voice sample and policy reference that informed it ## The edge ORZA does not invent brand voice. The voice baseline is frozen at deployment from the samples your CX lead has approved. When the brand voice needs to evolve, your CX lead changes the samples — not the agent. ## The hook Every Friday ORZA ships a brand-voice drift report: which responses tested closest to the baseline, which drifted, which received low customer ratings. The agent's own quality assurance, in plain language. ## Tools and integrations - Slack - Twilio - Vonage - Intercom - Zendesk ## Evidence - Operates under a four-tier constitution; brand-voice drift, refund commitments, and channel additions always require approval - Flat $1,000/month — no outcome-based negotiation, no failed-escalation upcharge - EvalOps suite covers brand-voice fidelity, escalation accuracy, and false-confidence rate - Inherits new base models and new agent skills automatically — pushed to every Fidelic agent the moment they ship. Same shape as a SaaS update; no upgrade purchase, no version pinning. ## Safeguards ORZA will not improvise brand voice. The voice baseline is frozen at deployment; only your CX lead can change it. ORZA will not commit to refunds, credits, or SLA exceptions; those always escalate. ORZA will not handle billing disputes, account closures, or legal-flagged cases. ORZA will not respond on a new channel without your CX lead's approval. ## Worst-case behavior ORZA pauses and escalates rather than guessing. Every escalation includes the trigger that fired and the rule that gated it; the conversation transcript follows. ## Day 1 / Week 1 / Month 1 - **Day 1:** Provisioned. CX lead approves brand-voice samples and channel scope. First test conversations run in sandbox. - **Week 1:** First channel goes live under your CX lead's review. Brand-voice drift baseline established. Escalation routing confirmed. - **Month 1:** Multi-channel coverage active. First brand-voice drift digest ships weekly. Customer rating delta reported. ## 90-day success criterion Brand-voice fidelity score at the level your CX lead defines, sustained for three weeks, with no CSAT regression. ## Integrations / supervision required Voice provider (Twilio, Vonage). Chat/SMS surface (Intercom, Zendesk). Read access to your brand-voice samples and your published policy. Slack for digests and escalation routing. ## Resume **Background.** Voice and multi-channel CX is the most opaque category in AI agents — enterprise-only pricing, negotiated outcome SLAs, and almost no published constitutional gating. ORZA runs the opposite playbook: flat pricing, frozen brand-voice baseline, and a four-tier constitution gating every response. **Methodology.** - Reads brand-voice samples and the last 90 days of conversation history before going live on any channel - Routes against a four-tier constitution: autonomous within the brand-voice baseline, review-required on channel additions, escalate on refunds/SLA/policy interpretation, refuse on legal-flagged cases - Logs every response with the brand-voice sample and policy reference that informed it - Surfaces brand-voice drift weekly for your CX lead's review **Evals.** EvalOps suite covers brand-voice fidelity, escalation accuracy, false-confidence rate, and customer-rating impact. Suite must pass before any channel goes live. **Operating record.** First-cohort deployments scheduled June 2026. Operating-record metrics publish with the first quarterly cohort review. ## Compatible agents - [AERA-01 — AI Agent-Assist Copilot](https://fidelic.ai/agents/aera) — Real-time assist for human support agents - [KESA-01 — AI Support Resolver](https://fidelic.ai/agents/kesa) — Tier-1 ticket resolution with published limits - [KORA-01 — AI Customer Success Lead](https://fidelic.ai/agents/kora) — Renewal-risk early-warning ## Related honest questions - [What do I actually own if I cancel my AI agent tomorrow?](https://fidelic.ai/hard-questions/what-do-i-own-if-i-cancel) - [Is Fidelic just a wrapper around GPT?](https://fidelic.ai/hard-questions/wrapper-around-gpt) - [Will my team see this as me trying to replace them?](https://fidelic.ai/hard-questions/team-replacement-perception) --- Canonical: https://fidelic.ai/agents/orza