A major US pharmaceutical manufacturer was handling more than 300,000 adverse-event cases annually through a process that depended heavily on contact-center agents. The agents had to manage conversations, capture critical information and complete the work required to create cases. The volume made the cost of that manual effort increasingly visible.

At a glance

The problem was compounded by after-call work and paper-based source documents. When information was missed during the interaction, agents and quality teams could need to return to recordings. The workflow therefore carried effort after the conversation had ended, even though the relevant information had already been spoken.

The organization needed to improve productivity without changing the fundamental role of the contact center. The opportunity was to make the conversation itself a source of structured case information, reducing the amount of manual reconstruction required afterwards.

The shift

Airo developed contact-center automation around the client’s Genesys Automatic Call Distribution environment. Real-time voice-to-text converted conversations into text, while NLP identified relevant information. The extracted content could be used to auto-populate the CRM, create the case and generate downstream outputs including E2B-compliant XML.

The important change was not simply to introduce automation into an existing task. It was to redesign the path through the work so that predictable activity could follow a defined workflow and the people responsible for the process could spend more time on the cases, decisions and exceptions that required their attention.

Transcribe while the call is happening

Real-time voice-to-text captures the conversation without waiting for the agent to reconstruct it later.

Extract relevant information

NLP identifies information needed for pharmacovigilance case intake.

Reduce manual case creation

CRM auto-population reduces repetitive re-entry of information.

Support multilingual and structured output

The solution supports six languages and can generate PDF, Excel, Word and E2B-compliant XML outputs.

How it works

The caller-agent interaction is transcribed in real time. NLP processes the conversation and identifies information relevant to case creation. That information can be transferred into the CRM, while structured outputs can be prepared for downstream pharmacovigilance processing. The workflow therefore connects conversation, extraction, case creation and downstream output.

What changed operationally

The productivity opportunity comes from reducing the distance between what is said and what becomes a case record. Instead of asking agents to perform every documentation step manually, the automation absorbs the repetitive information-handling layer while the agent remains part of the interaction.

That distinction matters because operational value compounds when the same workflow is executed repeatedly. A more structured process can reduce the amount of effort spent locating information, repeating predictable steps or maintaining visibility manually. It also gives teams a clearer place to intervene when something falls outside the standard path.

The approach is therefore less about removing people from the process and more about changing what people spend their time doing. Where the work can be standardized, automation takes the repeatable path. Where judgement is required, the workflow surfaces the work for human attention. That is what makes the model useful beyond a single automation task.

Why the operating model matters

The case demonstrates that automation becomes more valuable when it is embedded in the business workflow around it. A tool can complete a task, but an operating model determines how the task connects to what happens before and after it. In these engagements, the objective is to create a clearer sequence from input to action to outcome, with fewer unnecessary handoffs and less repetitive coordination.

This also makes the solution easier to connect to business ownership. Teams can see what enters the process, what the automation handles, where human review remains necessary and what outcome the workflow is intended to influence. That makes the improvement measurable and gives the organization a basis for extending the approach as requirements evolve.

The results are deliberately presented after the business problem and operating change. This keeps the focus on what was changed first, then shows what that change delivered.

The published case title reports a 70% improvement in overall productivity.

The organization was handling more than 300,000 adverse-event cases annually.

The solution supports English, Spanish, German, French, Italian and Portuguese.

Real-time capture reduces manual documentation and review effort.

From process change to business impact

The case shows how contact-center automation can be more than transcription. By connecting speech recognition, information extraction and case creation, Airo created a more direct path from the safety conversation to a usable pharmacovigilance record. At high volumes, that change in the workflow can translate into meaningful productivity gains.

The broader lesson is straightforward: the strongest automation stories are not about how many tasks a system can perform. They are about what becomes possible when a repetitive part of the operating model is made more consistent. That can mean faster turnaround, lower manual effort, greater capacity, better control or a clearer path to scale. The case demonstrates that connection without extending the published evidence beyond what the source supports.