Stonly Launches Business Process Agents for CX Automation

What happened

Stonly has launched Business Process Agents, a new category of AI agents designed to automate complex, multi-step customer service processes rather than simply answering discrete queries. The announcement, carried via PR Newswire, positions the launch as an extension of Stonly's existing guided-support platform into fuller process automation for support teams.

According to the announcement, the new agents are built to handle workflows that span multiple steps, systems or decision points — the kind of processes that have traditionally required a human agent to follow internal procedures, pull information from several sources, and execute actions on the customer's behalf.

Why it matters

This launch reflects a broader shift in AI-powered customer service: away from single-turn chatbots that answer FAQs, and toward agents that can execute end-to-end processes. For support organisations, that distinction matters because most costly, slow interactions are not simple questions — they are multi-step cases (a return, a billing dispute, an account change) that involve several systems and rules.

If Business Process Agents perform as described, they change what teams can hand off to automation: not just information retrieval, but the orchestration of a full case from intake to resolution. That has implications for staffing models, average handling time and where human agents are deployed — likely toward exceptions and judgment calls rather than routine procedure-following.

The Renascence take

The industry narrative around AI agents has moved fast from "answering questions" to "doing work," and this launch is another data point in that shift. The real test is not whether an agent can follow a process, but whether it fails gracefully and transparently when the process doesn't fit the customer's situation.

Most vendors sell automation on speed and cost savings; the metric that actually protects the brand is failure behaviour. A process agent that quietly does the wrong thing at step four of six is far more damaging than a chatbot that can't answer a question — because the customer often won't notice until the consequence lands. Before scaling any process-automation agent, operators should map where in the workflow human judgment is genuinely required, instrument the agent to flag uncertainty rather than push through it, and measure success by resolution quality and trust retained, not just tasks completed.