Sierra AI: Redefining Customer Experience in the Age of Conversational AI
How are AI agents revolutionizing customer experience? How Sierra AI is redefining what’s possible, and how will it shape the future of customer interactions?
Founding Story
Source: Wired.com
Sierra is a platform that enables businesses to build their own branded customer-facing AI agents. Founded in February 2023 by Bret Taylor (ex-Salesforce CEO and current advisory board chair at OpenAI) and Clay Bavor (ex-Google product leader responsible for Google’s AR/VR bets) this 1-year-old company is valued at $4.5B from its most recent fundraising round led by Greenoaks Capital.
Sierra was built on their shared vision to transform how businesses interact with costumers. Throughout history, advances in technology have triggered profound shifts in how businesses connect with their customers.
The Internet Era: Early websites served as digital business cards, offering little more than contact details or directions to physical stores.
The Social Media Revolution: Platforms enabled brands to engage directly with consumers in dynamic, two-way conversations.
The Mobile-First Wave: With the rise of smartphones, mobile apps became the primary interface between businesses and their customers.
Today, we stand at the cusp of another transformation—powered by conversational AI. This technology is poised to reshape human-computer interaction entirely, making natural, intuitive conversations the heart of every digital experience. Taylor and Bavor envisioned Sierra as the platform to drive this shift, enabling businesses to harness the full potential of this groundbreaking technology.
Customer Experience Revolution: Fixing a Broken System
Customer experience is broken. I can almost guarantee that you have had at least one bad customer experience - whether it is a cancelling a flight, returning a pair of shoes, or waiting hours on a call to get more information about different insurance plans. Customer service has been a universally bad experience. In fact, an average American waits 13 hours annually to get assistance from customer support agents and 65% of negative service experiences are attributed to slow case resolution. This can be frustrating and often leads to churn for businesses.
On the enterprise end, an average customer service call costs $13 which quickly becomes a big cost consideration as the customer service tickets scale with growth. There have been attempts to automate customer experience through the introduction of chatbots; however, existing chatbots offer standardized chats based on complex decision trees for common customer problems with significant upkeep costs. Even if there are only slight nuances involved, human agents need to be involved in the loop, which adds to mistakes, latency, and overall a poor experience. Sierra, through its branded AI agents, aims to provide delightful customer experiences while at the same time improve business outcomes.
Product
Sierra’s value proposition is simple: it provides a platform for companies to create branded customer-facing AI agents. Sierra’s agents offer delightful conversational experiences for their customer’s customers (end users of B2C enterprises) across customer support, sales assistance, subscription management, and more. Each Sierra agent can be tailored to reflect a brand’s unique voice and personality like Chubbies' youthful, witty "Duncan Smothers" or a refined tone for luxury brands. What’s also really interesting is that Sierra’s agents go beyond the conversational layer - they can take actions like tracking orders without human help, ultimately driving better business outcomes for the enterprises.
Source: https://sierra.ai/blog/introducing-sierra
To achieve this, Sierra provides an end-to-end platform supporting every stage of the AI agent development lifecycle. Its two flagship platforms—AgentOS and Experience Manager—make it seamless for businesses to design, deploy, and optimize production-grade branded AI agents.
Source: https://sierra.ai/blog/agent-development-life-cycle
1.AgentOS
AgentOS is the overarching platform that helps manage the development and deployment of AI agents and ensures they are are reliable, testable, and production-ready. It consists of 3 core components: AgentSDK, ExperienceSDK, and ContactCenterSDK.
AgentSDK
The Sierra Agent SDK is a core product offering as part of the AgentOS ecosystem. It that enables developers to build agents using declarative programming language. AI models are largely non-deterministic and agentSDK equips developers with tools to set deterministic guardrails that the agent cannot cross (e.g., orders can only be returned within 30 days of purchase.) This is extremely critical for production grade agents so mishaps such as the Air Canada incident are avoided.
The AgentSDK’s true power lies in its ability to abstract the complexities of agent development. Developers can focus on defining what behaviors and guardrails the agent should have, without worrying about the underlying AI models. This approach allows Sierra customers to build complex, yet maintainable and modular agents while entirely eliminating the chaotic workflows often associated with prompt engineering.
Also, The SDK orchestrates agent control flow across multiple systems, knowledge sources, and AI models. This enables agents to make contextual decisions, take appropriate actions, and respond meaningfully. In some cases, Sierra agents can fully resolve over 90% of customer inquiries, even in complex customer journeys.
Other highlights of AgentSDK include:
Immutable Agent Releases: Each release bundles source code, model versions, and knowledge snapshots, enabling instant rollbacks and precise A/B testing of new behaviors.
Seamless Upgrades: Agents automatically benefit from new AI models, like GPT-4, without requiring code changes.
Overall, AgentSDK delivers a robust product experience by empowering developers to build sophisticated, reliable, and easily maintainable AI agents while abstracting the complexities of underlying AI models.
Source: https://sierra.ai/blog/meet-the-ai-agent-engineer
ExperienceSDK
Experience SDK allows the enterprise to deploy AI agents across various channels. Through this feature, Sierra’s AgentOS goes above and beyond to make the release process delightful for enterprises by making it possible to build their company's AI agent once and deploy it on any channel - whether it is text or voice with only small modifications. Sierra’s "build once, deploy anywhere" model simplifies operations, ensuring updates—like new policies or product launches—are instantly reflected across all channels.
ContactCenterSDK
In cases where AI agents need to escalate issues, this SDK ensures a smooth handoff to human agents. By providing human agents with detailed AI-generated summaries, Sierra ensures consistent, high-quality customer experiences even during escalations.
2.Experience Manager
Experience Manager is a platform for observability, evaluation, reporting, and optimization of AI agents. It is built for CX teams so that they can monitor agent interactions in real-time and access detailed performance metrics such as resolution rates and customer satisfaction scores.
To help businesses understand and optimize their AI agents' performance in real-time, Sierra provides robust analytics solutions through its Experience Manager. Every customer interaction annotated in the Experience Manager feeds directly into Sierra’s improvement cycle. Also, since the agents are built in AgentOS, developers can trace the reasoning behind agent decisions, identify issues, and simulate problem conversations for rapid fixes.
Source: https://sierra.ai/blog/shipping-and-scaling-ai-agents
Testing
The annotations of various conversations also form the foundation for regression tests, almost guaranteeing that agents never repeat the same error. Sierra enhances this process with a simulation-based testing approach. Sierra uses a simulation-based approach to test its AI agents, pairing them with user simulators that mimic real-world behaviors and diverse user personas. A mock database validates whether the agent's tools are used correctly, enabling deterministic testing. This rigorous framework, combined with annotated regression tests, guarantees continuous improvement and production-grade reliability.
Overall, Sierra provides a very comprehensive platform and delightful experience for enterprises to build, test, and release product-grade branded AI agents.
Additional Notes
Sierra assigns an agent engineer and product manager to each client, ensuring businesses can deploy branded and customized AI systems in the easier way possible; thus reducing friction for adoption from potential customers. While this is not necessarily a ‘product feature’, this is a really crucial part of the process that leads to the development of the best agents.
Bavor, in his chat with Sequoia, mentioned a really interesting insight, “the solution to many problems with AI is more AI.” This stemmed from how Sierra employs supervisory agents to monitor the performance of the primary agent in real-time. These layers help ensure factual accuracy, compliance with business rules, and safe handling of sensitive information.
Business Model
Sierra AI uses an outcome-as-a-service model, marking a significant shift from traditional subscription based pricing and seat based pricing used by their cloud predecessors. Under this model, companies pay only when Sierra's AI agents successfully resolve customer issues or achieve desired outcomes, directly aligning Sierra’s success with that of its clients.
As Sierra's AI capabilities scale to handle more complex tasks and decision-making, the company stands to benefit from increased successful outcomes per customer and further improving their unit economics. Through this outcome-as-a-service model, Sierra AI is positioning itself not just as a product provider but as a partner in delivering measurable results, potentially revolutionizing how AI services are valued and priced in the enterprise market.
Market
Market Size
Sierra, through its agents, is offering service-as-a-software. That means the addressable market is not the software market, but the services market, which significantly expands its TAM.
Source: https://www.sequoiacap.com/article/generative-ais-act-o1/
As of November 2024, Sierra’s core product is providing customer support AI agents as a service. In the US alone, there are approximately 2.9M customer service representatives who earn an average salary of $40,000 per year, which leads to an approximate market size of $116B in the US. Since these AI agents don't have any language barriers and can operate 24/7, the market is global, which significantly expands its TAM. However, these are high-level estimates, and several nuances could affect the true market size. For instance, Sierra’s outcome-based pricing model—charging for resolved issues rather than per seat—differs significantly from traditional models. This shift not only redefines how value is measured but could also substantially alter market-sizing calculations, depending on client adoption and the complexity of tasks handled by AI agents.
Competitive Landscape
Overall, the AI customer support market is highly competitive with a lot of competition from both incumbent players such as Salesforce building Agentforce and startups such as Decagon building AI agents. For this analysis, I’ve focused on direct competitors—companies creating AI agents—categorizing them by the communication channels they support (e.g., voice, messaging, chat, and email).
The landscape is very dynamic and I hypothesize that the market is likely going to consolidate overtime and all competing firms will provide omnichannel support.
While the field is very competitive, there is reason to believe that Sierra will emerge as one of the winners if it continues building on the following moats:
Fast Shipping Velocity: Within just a year, Sierra launched AI agents capable of handling both text- and voice-based requests, outpacing competitors focused solely on text. This high shipping velocity gives Sierra a critical edge in product differentiation within a fast-evolving market where incumbents have a strong distribution advantage.
Technical Advantage: Sierra has a decided research team which I believe gives them a talent moat and technical moat in this space. In fact, they recently published advancements on measuring AI agent performance among other key findings which they implemented in their own product. If Sierra continues to foster the research and engineering collaboration, it could give the firm a strong technical advantage.
Strong GTM Motion: Sierra already has top consumer brands as their clients such as AG1 and SiriusXM etc. In this hyper-competitive market, the strength of the GTM motion is going to be as critical as product innovation and Sierra must keep its GTM momentum going strong.
Customers
Sierra is targeting B2C enterprises which seems unconventional as there is a lot of noise in the B2B agent space across other verticals like sales and marketing. However, in this case, Sierra’s decision to focus on B2C seems reasonable for mainly 2 reasons:
Higher Contact Volume: B2C brands handle significantly higher volumes of customer interactions compared to B2B enterprises, which typically serve a smaller set of key clients. This creates intense pain points for B2C companies, such as scaling support during peak demand and ensuring consistent, high-quality experiences across interactions. Sierra’s AI agents are uniquely positioned to address these challenges at scale.
B2C brands typically have shorted feedback cycles: B2C enterprises benefit from frequent, diverse customer interactions, which lead to shorter feedback cycles. This allows Sierra’s agents to learn and improve quickly, enhancing their ability to resolve a higher percentage of tickets successfully. Faster iteration and refinement help Sierra rapidly optimize performance, driving value for its clients.
Early Traction
Sierra has quickly gained traction with high-profile clients within the B2C space across various industries, including major brands such as Casper, Sonos, SiriusXM, and WeightWatchers. These clients have already witnessed substantial improvements in customer satisfaction and efficiency by integrating Sierra’s AI agents into their workflows.
For instance, at Casper, Sierra has helped increase the customer satisfaction (CSAT) score by over 20% and raised resolution rates by 74%, contributing to a more seamless post-purchase experience that fosters long-term customer relationships. Similarly, WeightWatchers has leveraged Sierra’s technology to engage members empathetically and at scale, achieving a resolution rate of nearly 70% and consistently high CSAT scores. This level of impact highlights Sierra’s capability to enhance customer experience, drive loyalty, and reduce support costs across a diverse set of industries.
Team
Sierra’s team is its strongest asset. It boasts an exceptionally strong founding team with a proven track records in building and scaling some of the most influential technology companies and products.
Bret Taylor, Co-Founder and CEO: Bret brings a wealth of experience from his roles at some of the most influential tech companies. He previously served as co-CEO of Salesforce, was the CTO of Facebook, and co-created Google Maps. His experience in leading large-scale technology companies and understanding enterprise needs is invaluable for Sierra's mission to transform customer service through AI.
Clay Bavor, Co-Founder: Clay's 18-year tenure at Google, where he most recently led Google Labs, provides Sierra with deep expertise in cutting-edge technology development. His experience in leading Google's AR/VR efforts, Project Starline, and Google Lens demonstrates his ability to drive innovation in emerging technologies, which is crucial for Sierra's AI-driven approach.
The founding team's extensive network and reputation in the tech industry have likely contributed to Sierra's ability to secure significant funding, attract high-profile clients like WeightWatchers, Sonos, and Sirius XM Holdings within a short period, and attract top talent. Their combined experience in enterprise software, AI, and customer-facing technologies positions Sierra strongly in the competitive AI agent market.
Key Opportunities
Multi-Modal Expansion: Sierra already provides voice and text-based solutions, but the next frontier lies in offering avatar-based AI agents. Avatars add a visual layer to interactions, making customer experiences more engaging and human-like like a personal brand ambassador.
International Expansion: With AI’s inherent advantages - 24/7 availability and multilingual support—Sierra is primed to support companies with global operations. As businesses increase their international reach, Sierra can offer robust, language-agnostic solutions to meet diverse customer needs.
Enhanced Personalization for Higher ARPU: Sierra has the potential to evolve its AI agents into true “personal concierges” by refining its models to deliver hyper-personalized interactions. For example, agents could proactively recommend tailored product bundles, anticipate customer needs based on behavioral data, or offer bespoke loyalty rewards. These deeper, more personalized interactions would not only enhance user satisfaction but also drive higher customer lifetime value and ARPU for Sierra’s clients.
Expansion into Proactive Customer Engagement: While Sierra currently focuses on reactive support, proactive engagement represents a significant growth opportunity. AI agents could identify potential issues before they escalate, such as warning customers about expiring warranties, delayed shipments, or unused subscriptions. This proactive approach would elevate the customer experience, reduce churn, and create additional revenue streams for Sierra's clients
Key Risks
AI Hallucinations: There is a risk that there are not significant improvements in agentic architectures and that AI hallucinations don't decrease significantly over time. Persistent hallucinations could undermine Sierra’s value proposition, especially in industries like finance, healthcare, or legal services, where precision and trust are critical. This issue could lead to decreased client adoption and increased reliance on human intervention, diminishing the cost savings and efficiency gains Sierra offers. However, with the talent and capital flooding to improving agentic architectures and solve for this core problem of hallucinations, this seems unlikely.
Build vs Buy: Companies like Klarna have started developing in-house AI agents, raising the question of whether enterprises will choose to build rather than buy solutions like Sierra’s. However, this scenario is unlikely to become widespread due to the high total cost of ownership (TCO) and complexity of building and maintaining production-grade AI agents. Sierra’s expertise, rapid iteration capabilities, and outcome-as-a-service model provide compelling reasons for enterprises to adopt rather than build.
Evolving Regulatory Landscape: The legal and regulatory frameworks governing AI are still in their infancy, and new laws could impose unforeseen compliance challenges. Sierra must remain agile, proactively addressing emerging requirements like data privacy, ethical AI usage, and liability concerns. Building a compliance-focused architecture could turn this challenge into a competitive advantage.
Competitive Noise and Performance Risks: The AI-driven customer experience space is highly competitive, with incumbents and startups alike vying for leadership. Sierra must continually innovate and ensure its AI agents outperform competitors to maintain differentiation. Any decline in agent performance or slower product updates could erode its position in this crowded market.
Growth Strain Risks: Rapid growth can strain resources and create operational backlogs. A fast-paced expansion might require Sierra to recalibrate its growth trajectory to manage resources effectively and sustain quality.
Financial Market Constraints: In a tightening financing market, Sierra could be forced to rely more heavily on cash flow over external funding. This scenario may limit growth initiatives and necessitate a more conservative financial approach.
Summary
Sierra AI operates in a highly competitive market of LLM-powered customer service solutions. Its platform offers features like branded AI agents, omnichannel support, real-time analytics, and advanced testing frameworks, though similar capabilities exist among competitors. Sierra’s edge lies in its rapid product iteration, exceptional founding team, and strong technical foundation. With a massive market opportunity and the continuous advancement of agentic architectures to reduce hallucinations, Sierra is well-positioned to capitalize on this growth. Despite challenges from competition and evolving regulations, Sierra is poised to become a leader in reliable, scalable AI for customer support.
Well written, very thoughtful and well researched.