Cresta, Which Serves Marriott, Says Its Agent Builder Roughly Halves Deployment Time

Cresta, the customer experience AI vendor whose enterprise clients include Marriott, says its new agent builder cut initial deployment time roughly in half in early implementations. The detail comes from Anthropic’s Claude Blog, which describes Conductor, a meta-agent built on the Claude Agent SDK that guides a team from initial blueprint through implementation, evaluation and optimization. Cresta, founded in 2017, reports more than $100M in ARR and has raised more than $270M.

For hotel operators the number matters because the slow part of guest-facing AI has been the build cycle, not the model. Conductor helps developers decide which workflows need deterministic control and which can stay flexible. In a hotel that split is concrete: a rate-integrity or payment step should follow a fixed path, while a guest asking about late checkout can be handled with room to adapt. Cresta also captures proven patterns as reusable skills, so a pattern that works at one property does not have to be rebuilt for the next.

Faster deployment raises the stakes on the handoff to a person. A Hospitality Net opinion piece cites survey data showing 84% of U.S. customers say AI-powered customer service should always include easy access to a human, while 71% say AI makes it harder to reach one. An agent that ships in half the time and traps guests in a loop is a faster way to damage review scores.

The operator question for any vendor pitch like this one: how is the agent scored before it touches a guest? Cresta evaluates Conductor on creating agents, writing test cases, modifying agents and root cause analysis, with scoring that includes whether the execution path was appropriate. Ask your vendor for the equivalent, and ask where the human escalation path sits in the blueprint.