Cresta Self-Service AI Agent Platform Opens to Developers

Cresta has opened its customer experience AI platform to full developer self-service, giving enterprise teams the tools to build, test, deploy, and optimize customer-facing AI agents without waiting on a vendor deployment team. The Cresta Self-Service AI Agent Platform, announced on October 29, 2026, centers on Cresta Conductor, a deployment system that lets developers work from their own code repositories and continuous integration pipelines. According to the announcement, some enterprise developers have already taken agents from first requirements to production in as little as one week using the new workflow.

The move marks a shift in how Cresta delivers its technology. Until now, the Sunnyvale company deployed AI agents for customers largely through its own forward-deployed engineering teams. The Cresta Self-Service AI Agent Platform hands those same tools to customer and partner developers, a change that reflects a broader trend in enterprise software: engineering teams increasingly expect AI agents to move through the same code review, testing, and release discipline as any other production system, according to the company.

Cresta describes itself as an AI operating system for customer experience, where human and AI agents work together across the customer lifecycle. Its customers include United Airlines, CVS Health, and Marriott, according to the announcement. The self-service opening therefore matters beyond a single product update: it tests whether the practices Cresta refined while deploying agents for large enterprises can survive contact with in-house teams that have their own tooling, timelines, and standards.

What the Cresta Self-Service AI Agent Platform gives developers

At the center of the Cresta Self-Service AI Agent Platform sits Conductor, which was built on the Claude Agent SDK in collaboration with Anthropic. Conductor was originally developed to help Cresta's own deployment teams build agents faster, and according to the company, those teams now spend up to five times less time on custom tools, code, and deployments. Customer and partner developers now get the same Conductor tooling plus the Cresta CLI and the Cresta AI Agent SDK for writing custom code in standard Python that connects agents to internal systems.

Developers work from their own terminals and coding assistants rather than a separate visual builder. Because Conductor supports the open Agent Skills standard, skills a team already wrote for its coding assistants carry over instead of being rebuilt. That design choice addresses a common frustration with agent platforms: work done in one tool often stays trapped there. By meeting developers inside their existing repositories and pipelines, the Cresta Self-Service AI Agent Platform lowers the cost of switching and experimenting.

One example cited in the announcement illustrates the pace Cresta is promising. A developer at an Australian telecom provider rebuilt an existing production chat agent, with more than ten subagents and custom code, from another platform onto Cresta in under a week. Early adopters at global service providers report similar momentum: Atento plans to build an AI agent workforce for its clients on the platform, while TELUS Digital engineers have been building on Conductor since June, according to the announcement. Those figures come from the vendor and its customers, not independent audits, so they describe early outcomes rather than guaranteed results.

Why self-service changes the agent build

The case for self-service rests on a simple observation reported by Cresta: many engineering teams already build software with coding agents, and they expect agent projects to follow the same process. When agents live only in a visual interface outside that process, version control, testing, and staged releases become afterthoughts. The Cresta Self-Service AI Agent Platform instead treats an agent as software, with discovery, build, testing, go-live, and continuous improvement handled on the team's own timeline.

That framing connects to a wider pattern across the agent ecosystem this month. Enterprises are moving agents from pilots into production roles where governance matters as much as capability, a theme also visible in recent launches of governed agent browsers and agentic IT support tools. The difference with Cresta is the audience: rather than selling agents as a finished service, the company is selling the factory that builds them, betting that in-house teams would rather own the process than rent the outcome.

Self-service also changes who can build. According to the announcement, any team that writes and ships software can build production-grade agents on the platform without a dedicated team of AI specialists. That claim will be tested by teams whose first agent is also their first encounter with evaluation, guardrails, and release management for conversational systems. Cresta Academy, a new training and certification program, is designed to close that gap by teaching engineers to design, test, launch, and improve agents, and by giving technology leaders a standard for which engineers are ready to ship on their own.

Trust controls for agents facing customers

Putting agents in front of customers raises the stakes for every release, and the Cresta Self-Service AI Agent Platform pairs its builder tools with production controls. Teams can send a share of conversations to a new agent version, run A and B tests between versions, and roll back when a version underperforms. Builder and reviewer permissions set who can build, review, test, and inspect each agent, separating the roles that create an agent from the roles that approve it.

Compliance credentials back those controls. According to the announcement, Cresta holds SOC 2 Type II, ISO 27001, and ISO 27701 certifications, is PCI DSS and HIPAA compliant, and was the first customer experience AI provider to achieve ISO 42001, the international standard for AI management systems. For regulated industries such as health care and financial services, that compliance posture is often a precondition for letting an agent handle customer conversations at all.

Visibility into usage and cost rounds out the platform. Teams get infrastructure for controlled releases, access control, and insight into AI usage and spending, according to the announcement. Cost visibility matters because agent workloads can scale unpredictably with conversation volume, and a platform that hides that meter until the invoice arrives will lose the trust of the finance teams who approve it.

What to watch next

The open questions around the Cresta Self-Service AI Agent Platform are practical ones. Pricing was not disclosed in the announcement, and the strongest evidence so far comes from early adopters already working closely with Cresta, including Atento and TELUS Digital. Independent results from teams adopting the platform cold, without that hands-on history, will be a better measure of whether self-service truly shortens the path to production.

Competition is also intensifying. Agent builders, customer experience vendors, and model providers are all racing to own the layer where agents get built and governed, and the open Agent Skills standard may help developers avoid lock-in across those options. Cresta's bet is that its deployment experience with large enterprises, productized into Conductor and its SDKs, is a durable advantage that a generic framework cannot easily copy.

For now, the announcement signals a maturing market. The first wave of enterprise AI agents was built for customers by vendors; the next wave, if Cresta is right, will be built by customers themselves, with vendor platforms supplying the rails. Readers following the agent platform race can compare this launch with recent governed agent launches covered here, including the Island enterprise browser for agents, ConnectWise agentic IT support agents, and Tencent WorkBuddy agent growth. Sources for this article include the company announcement carried by PR Newswire, coverage by CMSWire, and analysis by TipRanks.