← ALL PROJECTS
InsightBoard
An analytics workspace that turns dense operational data into clear, useful views a team can actually act on — without needing a data team on call.
Overview
Operational data, made readable for everyone.
InsightBoard started from a familiar frustration: operations teams sitting on piles of exports and dashboards nobody opens. I set out to build one calm workspace where a team can connect a data source, ask a plain-language question, and get a chart plus a short written takeaway.
The front end is a Next.js workspace with saved views and shareable links. A Python service prepares and caches the datasets, and an OpenAI-assisted layer drafts the summaries and suggests follow-up questions — always beside the raw numbers, never instead of them.
Approach
How the build came together.
- 1Mapped the questions first. Interviewed three operations-minded classmates and listed the ten questions they actually ask every week — the workspace answers those before anything else.
- 2Kept the pipeline boring. Nightly Python jobs clean and cache each dataset, so the UI stays fast and every chart can show exactly when its data was refreshed.
- 3Summaries with receipts. Every AI-written takeaway links back to the rows and filters behind it, so a reader can verify a claim in one click.
- 4Designed for sharing. Saved views get stable links with sensible defaults, so a chart can travel from the workspace into a team update unchanged.
Gallery
Inside the workspace.
Related work