A plain-English interface still leaves an engineer in the path if the reporting system cannot supply the answers. The reporting system has to be able to answer before that interface is worth building.
At a US consumer app, a business question often turned into an engineering task. Someone had to filter the data, prepare an export, and plot the result. Reports took hours or days. Decisions waited, and engineering work was interrupted for work that was really reporting.
The constraint was straightforward. The work had to support nontechnical team leads, and it had to connect to live business data. An easier interface would only be useful if the underlying reporting system could supply the answers.
I started with that system, not with the language interface. I built a near-real-time analytics stack with ClickHouse and Apache Superset, including KPI dashboards and scheduled email reports. The work included internal infrastructure and a Superset deployment on AWS Fargate. That foundation came first.
I then added the official Superset MCP service, which is part of the Apache Superset codebase, so nontechnical team leads could ask questions of live business data in plain English. The dashboards and the language interface shared the same purpose: make business data accessible without an engineer preparing each answer. Routine reporting moved from hours or days to seconds. Team leads could answer questions themselves.
The useful lesson is the order of the work. The connector came after the analytics stack and sat on top of it. The case study's constraint says why: an easier interface would only be useful if the underlying reporting system could supply the answers. If that system is missing, a well-worded question still ends up with an engineer preparing the answer.
Natural-language access also adds another interpretation layer. Clear metric definitions, query complexity, and source freshness still determine whether an answer is useful. A confident sentence can be wrong if the metric is ambiguous, the query is too heavy, or the data is stale. The analytics foundation remains essential even when the interface becomes conversational.
Dashboards and scheduled email reports already answered routine questions without an engineer preparing each one. The MCP connector served the same purpose for nontechnical team leads, in plain English, on the same live business data.
The case study for this work is From a reporting bottleneck to business answers.
If a reporting bottleneck like this is sitting in your business, I take that work through Developer Labs.