Serving customer experience and engineering teams—all from one platform

Quite soon, almost every business will have an AI agent, just like they have a website or app today. These agents—sophisticated software that can complete tasks unsupervised—will be the primary way customers engage with brands, and vice versa. In fact, agents built on Sierra are already handling everything from recommendations and discovery, to shopping, subscription management, and customer support. And we’re only just getting started.

Agents are the product of many teams, not just one

Over the last year, working with hundreds of companies, we’ve learned that agents aren’t the product of one team. Successful deployments are built on a close collaboration between customer experience, operations, engineering, and marketing—and we’ve designed Sierra to support that cross-team partnership. Yet almost every other solution available on the market today forces businesses to optimize for one group or another.

Build your own vs no code: a false choice

Current solutions fall into two distinct camps. On one end of the spectrum are the frameworks which enable engineering teams to build agents from scratch using today’s large language models. This approach has several drawbacks. First, it excludes the customer experience and operations teams. Second, engineering has to build everything—and while rolling your own agent sounds exciting, it’s much harder than people imagine. We know from personal experience! Third, even the smallest changes take engineering support.

At the other end of the spectrum are the no code tools that make it possible to build more basic agents quickly and maintain them easily without engineering support. But these struggle to handle complex workflows, enforce sophisticated guardrails (important in regulated industries) or integrate with legacy back-end systems—all requirements if agents are to personalize conversations and take action. Nor can they be managed as part of a company’s existing software development cycle.

The Sierra solution: one platform, multiple ways to build

Sierra’s Agent OS is the only platform that supports both no code and programmatic agent development. Our unique approach enables companies to mix and match—choosing the right approach for their business needs, and scaling their agent development as those needs evolve.

Agent SDK

Agent SDK is Sierra’s platform as a service for building agents with code. It enables any developer—from experienced engineers to generalists with no expertise in AI—to define their own customer journeys, the workflows which describe what the agent should do, and how. The goals and guardrails set for these journeys can be really simple or highly sophisticated, but either way, the agent will figure out how best to implement them. Our Agent SDK also fits seamlessly within existing software development life cycles, with version control, release gating, continuous integration, delivery and deployment, audit, and much more.

Agent Studio

Agent Studio is Sierra’s no code tool. It enables any team to define customer journeys in plain English based on the existing procedures used by the customer experience and operations teams—there’s no need for historical conversations. It’s a great way for anyone to build and maintain agents without engineering support. Agent Studio also enables companies to create the right brand and tone for their agents, and ensure changes to their business (like new promotions or products) are automatically reflected in customer journeys so the agent can do its job effectively.

Here’s an example of this mix and match approach in practice. A retailer implements a warranty claim workflow in code using our Agent SDK so it can effectively enforce guardrails from other systems—such as the need to check uploaded images to determine whether a product is genuinely damaged. But a returns flow, which changes with new product launches, is better built and maintained by the customer experience team in Agent Studio without code or the need for engineering support.

Continuous improvement

Of course, building an agent is just the start. Every one of our customers (however their agent is built) has access to Sierra’s complete suite of tools to test, monitor, and continuously improve their agents' performance, proactively and at scale. These include the ability to:

• Run AI simulations to stress-test workflows and improve agents before letting them loose with customers;
• A/B test conversation designs (for example, the most effective way to ask for a specific piece of information) to understand what works best; and
• Evaluate real customer conversations using AI-powered analytics to identify where an agent succeeds or fails, and automatically surface opportunities for improvement.

Better platform + better partnership = better performance

A great platform is important, and it’s just one part of the story.

Over the last year, we’ve learned the importance of partnership not just between customer experience, operations, and engineering teams, but Sierra and the businesses we serve. Our forward-deployed agent development team—which we embed with engineering, customer experience, or operations, depending on a company’s needs—helps ensure the agents built on our platform are successful and deliver the outcomes we promise (and which customers pay for). Because what businesses care most about is not shipping agents but shipping success: higher customer satisfaction, lower churn, a stronger brand, increased revenue, and improving margins.

If you want to learn more about how Sierra Agent OS makes it easy to design, test, deploy, and improve agents at scale, with or without code, all from one platform, we’d love to hear from you.