Lean teams stay lean by protecting their depth. Helply gave Covidence’s support team back the hours they used to spend explaining the same workflow three times a day.
Covidence builds systematic-review software used by medical researchers worldwide, the kind of work where a stuck import or a misconfigured workflow costs a researcher their afternoon, sometimes their week. Their support tickets are dense, technical, and tied to deadlines that don’t move.
The team was built for that depth. Smart, technical, fluent in both the product and the research workflows it supports. But every team has a finite amount of attention, and as Covidence grew, more of theirs was getting consumed by the same recurring questions, workflow setup, plan limits, integration steps. Important to the customer asking. Repetitive at scale.
Razia Aliani, Senior Systematic Reviewer at Covidence, knew the team’s value lived in the harder questions. The challenge was protecting that capacity from the steady pull of routine tickets, without growing the team in a way that didn’t fit the company.
Covidence had seen generic chatbots before, the pattern was familiar: confident, fast, often wrong. For a researcher on deadline, a wrong answer is worse than no answer. Whatever Covidence chose had to clear a high bar from the first day in production.
Helply trains on the customer’s actual support history, the real questions, the answers that worked, the tone the team uses. By the end of the first month, the AI was steadily resolving routine workflow and setup questions. Steady-state, around 62% of inbound conversations. At peak, 70%. The variance came from volume, not accuracy, the agent handles routine questions consistently, regardless of how many arrive at once.
For a lean team supporting researchers on deadlines, the practical impact wasn’t a metric on a dashboard. It was time. Hours that used to go into rewriting the same explanation for the third time that week now go into the customer-success conversations the team was built for: helping a researcher think through a complex review setup, troubleshooting a difficult import, walking a new team through their first protocol.
Covidence is now joining the first cohort to migrate fully to the AI-native helpdesk, outcome-based pricing, no seat fees, helpdesk always free. The AI Agent proved the model works at Covidence’s depth. The next step is replacing the existing helpdesk infrastructure entirely, with AI Drafts on every non-resolved ticket, automatic feature-request routing, and Ask Helply across the full support history.
For a research-focused team, the loop closes here. Every researcher question, every workflow stumble, every feature request becomes part of a queryable knowledge base, the kind of operational intelligence that only emerges when support data is properly captured and accessible.
