Editorial independence: AI Agent Square is not paid by the vendors we review. We currently earn no commissions from links on this site, and no vendor can pay to influence scores, rankings, or review content. How reviews are made is on our methodology page.
Sierra AI Pricing (2026): Outcome-Based and Quote-Only
There is no public pricing, no self-serve plan, and no free tier. Sierra does not publish a rate card. Third-party estimates circulate but are not confirmed by Sierra, so we do not repeat them here.
Sierra describes its model as outcome-based: it gets paid when its agent completes a task, such as a resolved support conversation, a saved cancellation, an upsell or a cross-sell. Sierra states that when a conversation is unresolved, or a case is escalated, in most cases there is no charge. Routing or greeter-style interactions may instead be billed on a consumption basis, by conversation count. The per-outcome rate and the definition of a qualifying outcome are agreed per contract and are not disclosed publicly.
What We Like & What We Don't
What We Like
- Voice is a first-class channel, and Sierra announced a Level 1 PCI-compliant payment capability across chat and voice: callers key in card details and the agent steps out while the processor completes the payment.
- Outcome-based pricing structurally aligns Sierra with results: you are billed for resolved outcomes, and unresolved or escalated conversations generally carry no outcome fee.
- Safety model: supervisors that enforce guardrails, policies and quality checks, plus deterministic guardrails in the Agent SDK that the agent cannot cross.
- Platform breadth: Agent OS, an Agent SDK with declarative goals and composable skills, Agent Studio, and Insights and Explorer analytics.
- Ghostwriter builds agents for voice, chat and email in more than 30 languages from a natural-language description.
- Compliance badges on sierra.ai include SOC 2, ISO 27001, ISO 42001, HIPAA, GDPR and PCI DSS.
What We Don't
- No published pricing, no rate card, no self-serve path — every evaluation starts with a sales scoping call, which is friction for procurement teams doing early research.
- The quote-only enterprise sales motion is built for large enterprises, not SMBs or most mid-market organisations.
- Outcome pricing sounds simple but requires careful contract work — the definition of a "qualifying outcome" and the per-outcome rate are negotiated individually and can be hard to model in advance.
- Narrow focus: Sierra is a customer-experience platform, not a general enterprise agent for internal productivity or sales automation.
Detailed Feature Review
Sierra was co-founded in 2023 by Bret Taylor, former Co-CEO of Salesforce and later chair of OpenAI's board, and Clay Bavor, who spent 18 years at Google. Sierra is based in San Francisco.
In May 2026 Sierra raised a $950 million Series E led by Tiger Global and GV at a $15.8 billion post-money valuation, per CNBC, which puts total funding at about $1.5 billion. Sierra reported over $150 million in ARR in February 2026 and says it works with 40% of the Fortune 50. Customers it names include WeightWatchers, SiriusXM, ADT, Chime, Nubank, Ramp, Rivian, Rocket Mortgage and Sutter Health.
Agent OS: The Platform Layer
Agent OS is the layer on which every Sierra agent is built, deployed and supervised. Sierra says it assembles agents from 15+ frontier, open-weight and proprietary models, with routing and orchestration handled by the platform, and deploys a single agent across chat, SMS, WhatsApp, email, voice and ChatGPT.
Agent SDK and Agent Studio: How Agents Get Built
The Agent SDK lets developers write agent behaviour in a declarative language, organised into composable skills, alongside deterministic guardrails the agent cannot cross, in a way that is abstracted from the underlying LLMs. Sierra frames it as part of an agent development life cycle that runs from development through testing and release.
Agent Studio is the counterpart to the SDK for teams that build without code. In practice the split is between engineering-owned logic in the SDK and operations-owned experience design in Studio.
Multi-Model Orchestration
Rather than routing every interaction through a single model, Sierra picks models per task through task-specific evaluations, and fine-tunes its own where off-the-shelf models fall short. Sierra says this keeps agents running if a single model or provider goes down, and lets them improve as frontier models improve.
Buyers with strict model-governance requirements should ask during scoping which models handle which interactions.
Vendor-Reported Results
We have not run a controlled test of Sierra's agents, so we make no claim about how they compare with human agents. Sierra's own WeightWatchers case study reports that the agent contained nearly 70% of cases in its first week with a satisfaction score above 4.5 out of 5. Treat figures like that as vendor-reported rather than independently verified; they are a reason to run your own pilot, not a substitute for one.
Voice AI: The Clearest Differentiator
Sierra treats voice as a first-class channel. It has announced what it calls the first Level 1 PCI-compliant payment capability for AI agents, verified through the Visa Global Service Provider Registry: callers enter card details on the keypad, and the agent switches to a secure transaction flow while the processor completes the payment. That moves voice from "answer questions" to "complete regulated transactions".
Ghostwriter: Sierra's Agent for Building Agents
Ghostwriter is Sierra's agent for building agents: you describe the customer experience you want in natural language, and Ghostwriter builds an agent for voice, chat and email in more than 30 languages, with safety controls included by default. The point is time-to-value for standard scenarios.
Supervision, Guardrails and Trust
Alongside the agents that talk to customers, Sierra runs supervisors that enforce guardrails, policies and quality checks. On top of that sit deterministic guardrails in the SDK: hard business rules the agent cannot cross, such as only processing returns within 30 days of purchase. For regulated industries this layered model means compliance boundaries are enforced by rules you can audit, not merely by the model's judgment.
Compliance and Security
Sierra's site lists SOC 2, ISO 27001, ISO 42001, HIPAA, GDPR, PCI DSS and FedRAMP badges, and its payments announcement describes Level 1 PCI compliance. Ask for the audit reports and their scope during procurement.
Channels and Build Tools
Sierra does not publish a connector directory. Confirm your specific CRM, commerce and telephony stack during scoping.
Who Should Use Sierra AI — and Who Should Skip It
Best For
Sierra is built for large enterprises with high customer-interaction volumes where customer-experience quality is a strategic priority rather than a cost line. Organisations with meaningful phone volume get the most from its voice channel and PCI-compliant payments. If you can absorb a quote-only enterprise contract and a planned deployment, Sierra belongs on your shortlist.
Who Should Skip It
SMBs, startups and most mid-market organisations should not attempt to evaluate Sierra: the quote-only sales motion means you cannot even benchmark it quickly. If you need transparent, self-serve, per-resolution pricing and a fast setup, look at Intercom Fin or Tidio Lyro instead. Teams already standardised on a specific stack may prefer a native option such as Zendesk AI or Salesforce Agentforce.
Alternatives to Sierra AI
Verdict and Recommendation
Sierra leads on the capabilities that matter for enterprise CX: voice with PCI-compliant payment handling, supervisors and deterministic guardrails, an SDK-based build stack, and an outcome-based commercial model. This assessment rests on Sierra's documentation, announcements and news coverage, not a hands-on deployment.
Two cautions. Pricing transparency is poor: there is no rate card, no self-serve path, and per-outcome rates are negotiated case by case, which makes early evaluation hard. And the quote-only enterprise motion restricts Sierra to a narrow band of very large organisations.
Our recommendation: if you are a large enterprise with serious interaction volume, especially phone volume, put Sierra on your shortlist and insist on a pilot scoped against your real customer data, not a generic showcase. Benchmark it directly against Decagon, the closest enterprise rival. If you are below that tier, do not spend cycles on Sierra: Intercom Fin or Zendesk AI will get you to value faster.
Evaluate Sierra AI for Your Enterprise
Compare Sierra against its closest rivals and the broader customer-service AI field before you commit to a quote-only contract.
Sources
- Sierra home page (channels, products, customers, compliance badges), Sierra, checked 11 October 2026
- Outcome-based pricing for AI agents, Sierra, checked 11 October 2026
- Constellation of models, Sierra, checked 11 October 2026
- The agent development life cycle, Sierra, checked 11 October 2026
- Level 1 PCI-compliant payments for AI agents, Sierra, checked 11 October 2026
- Agents as a service (Ghostwriter), Sierra, checked 11 October 2026
- Year two in review, Sierra, checked 11 October 2026
- Bret Taylor's Sierra raises nearly $1 billion months after last capital push, CNBC, checked 11 October 2026