Political research takes three weeks. The news cycle takes three hours.
Precinct is an AI political scientist. It runs the literature review, assembles the data, states the finding, and shows you the uncertainty around it. Built for research organizations, newsrooms, and campaigns.
We will write once, when there is something to use. No list sharing, no newsletter.
Specimen output
How much of the suburban turnout shift is explained by registration change rather than persuasion?
Registration change accounts for most of the shift, but the estimate is wide enough that persuasion cannot be ruled out.
- Statewide voter file, 2018 to 2024
- Precinct-level certified returns
- Two national probability surveys
Precinct does not report a point estimate without the interval around it, and it names every source it used. Illustrative example.
What it does
Reads the field
Finds the relevant literature, prior surveys, and existing estimates, then tells you where they disagree and why. The disagreement is usually the finding.
Builds the dataset
Pulls returns, voter files, survey microdata, and administrative records into one analysis, and documents every join and every exclusion so someone else can check the work.
Shows its uncertainty
Reports intervals, not headlines. When the data cannot answer the question, it says so instead of producing a confident number.
Built for the people who have to be right: research organizations producing work that gets cited, newsrooms on a deadline that cannot afford a correction, and campaigns deciding where the next dollar goes.