Be your own investigative data desk
“Find the story in my data”
For journalists, researchers & investigators
Before Sidecar
Sample the data in a spreadsheet. Miss the patterns hidden in the full set. Write a thin story. Your editor asks for data to back the claims — you can’t provide it. The story runs without evidence, or doesn’t run at all.
The moment it clicks
A FOIA dump arrives — 500,000 rows of procurement data. Most AI says “that file’s too large.” Sidecar opens it and starts surfacing patterns in seconds.
With Sidecar
A real conversation — you ask, it works, results appear.
- 01Drag the file in. It loads. No size-limit conversation.
- 02“Find unusual patterns — contractors whose billing doubled year-over-year, agencies with an odd concentration of awards, anything that looks wrong.”
- 03It scans the full dataset and surfaces 12 flagged patterns.
- 04“Drill into #4 — the contractor with the 400% jump. Show me all their contracts by year, as a chart.”
- 05The chart renders. The data’s still loaded — nothing to reload.
- 06“Now compare them to their competitors in the same category.”
- 07Twenty minutes in, the story has verified data, charts, and a clear narrative.
Why this is impossible without execution
The data is too large to fit in a chat window. The analysis needs real statistical computation, not an approximation. And the back-and-forth — drilling in, comparing, cross-referencing — only works because the full dataset stays loaded across every question.
What you’d see
Output preview — a chart showing one contractor’s billing spiking 400% against a flat industry average, with a table of their ten largest contracts.