OpenAI killed the SQL backlog with a GPT-5 analyst

The bottleneck at OpenAI wasn’t a lack of data — millions of support tickets pile up every year — but the brutal economics of analyzing it. A product leader’s simple question about a new feature’s reception required a data scientist, a week of SQL queries, and a custom classifier. “The process required deep technical expertise, and it was cutting off our curiosity,” says Molly Jackman, Head of Business Data. That friction meant dashboards showed what was happening but almost never the ‘why,’ forcing teams to ration their analytical bandwidth.

The fix is an internal research assistant that marries structured dashboarding with a conversational GPT-5 layer. A product manager can pull up a chart of trending issues and then immediately ask a messy, plain-language follow-up: “What are healthcare customers saying about new integrations?” Within minutes, the system returns a report sizing the problem, showing prevalence, and highlighting specific friction points. The output isn’t just a summary; it’s a quantified map of user sentiment that took virtually zero technical lift to produce.

Trust had to be earned before the tool became a daily habit. Early on, ops teams ran manual classifications against the assistant’s output, and data scientists wrote custom models to verify its accuracy. The results aligned, and as leaders cross-checked findings against anecdotes from the field, confidence solidified. That cycle of ask, check, and trust collapsed a week-long analytical slog into a few clicks. The payoff is now visible across the company: after GPT-5 launched, product teams synthesized feedback themes in days instead of weeks; when enterprise adoption of connectors slowed, the assistant immediately surfaced a buggy onboarding flow as the root cause.

This isn’t replacing data scientists. It’s freeing them from one-off fire drills to build better classifiers and automation. Ops teams now generate launch reports in minutes instead of days, shifting their focus toward direct customer interaction. Jackman frames the ultimate win as “customer UX research at scale,” where the voice of the customer stops being a backlog graveyard and starts actively reshaping product roadmaps and engineering priorities in real time.