Five gates and four practical prioritizers for choosing an internal AI pilot that a small operating team can evaluate and maintain.
Blog
Original essays, project notes, talks, and reporting connected to our work.
A day-30 review for deciding whether an internal AI pilot has enough evidence to expand, repair, stop, or keep observing.
Jeremy Herzog’s State of the Map US talk on making a rescued infrastructure catalog durable and legible to people, software, and agents.
The launch record for a public catalog that makes more than 400 preserved HIFLD Open datasets searchable, previewable, and downloadable.
Rajan Desai on why useful internal AI begins when the people closest to the work can follow a question through company data without losing the evidence.
Atmika Pai describes a conversational restaurant map built with Marauders’ geospatial tools and shaped through collaboration with Fulton Ring.
Why restoring a public dataset requires access, context, maintenance, and a community prepared to keep it useful.
The Guardian’s interactive reporting on the Future Risk Index and the preserved data Fulton Ring used to restore public access.
Jeremy Herzog’s original case note on building a data lake and reporting system for high-volume retail transaction data.