Back to Inside the AI Startup Making Customer Support 65% Cheaper, with Jesse Zhang.
How Does Decagon's AI Customer Support Work?
Jesse Zhang says Decagon's customer support AI is built, tested, and monitored inside the product, so the customer owns the results and the process. People from Decagon still stay on the account, because an enterprise rollout is not a handoff of software to go use alone. The build is not code an engineer has to reconfigure, and not a change Decagon has to come back and make.
Sierra gets into enterprises and has scaled its team quickly. Sierra's deployment is closer to Palantir: forward-deployed engineers who tell the customer they can solve the problem. A do-it-for-you model fits a large company that lacks the talent to move fast. The same model makes less sense as AI models get better, because agentic AI lets people on the customer's own team do more. Decagon keeps the white-glove staffing and points the product at ownership. Where Salesforce and Google are the default tools, the deals come from a product the customer's team controls.