What happens when we show up.

The organization

A local hospital foundation with a team of eight people, responsible for raising $5.5 million annually. They faced an extraordinary new challenge: raising an additional $35 million for the construction of a new hospital, a goal that demanded far more donor research capacity than the team had ever needed before.

The problem

Researching a single potential major donor, gathering everything known about them and then entering it into the donor relationship management platform, took one person about seven and a half hours.

Only one person could do it, it was often deprioritized, and the pace of research could not keep up with the pace the campaign required.

The StilwaterAI approach

We began with an education session for the entire team on how to get accurate, reliable information from AI tools. We then facilitated a workshop to identify and prioritize potential AI-powered assistants, evaluating each against return on investment, productivity improvement, cost reduction, alignment with the foundation's direction, and build time.

The top priority became an assistant called Yukon Cornelius. It automates donor research and updates donor profiles directly in the foundation's donor management software, with one critical design decision built in: a human approval layer reviews all research before it is committed. AI accelerates the work; people remain accountable for the decisions.

The outcome

Donor research that took seven and a half hours now takes about ten minutes, at equal or higher quality.

The bottleneck is gone. Anyone on the team can do research at any moment, and knowledge is shared across the organization rather than trapped with one person.

The organization

A California-based staffing company with 70 employees, placing day labourers and running weekly payroll. Their differentiator: they accept timesheets in any format from any client. It is a promise that wins business, and creates an enormous operational challenge.

The problem

Up to 3,800 timesheets arrived every month through a central email inbox. Four payroll processors handled every one by hand: opening, reading, validating, and entering data.

The team spent about 60 hours a week on this work, and manual cut-and-paste carried a constant risk of error.

The StilwaterAI approach

We evaluated the full business process behind payroll, then built an AI-powered assistant called Buddy.

Buddy ingests every incoming timesheet from the inbox, cross-references it against the submissions expected that week, and routes it into the company's payroll system without manual handling.

The outcome

Of the 60 hours a week the processors spent on timesheets, about 40 have been reclaimed, and the risk of manual-entry errors is significantly lower.

In a demonstration, the team processed 94 timecards in a single 20-minute session.

The organization

Us. To build thought leadership, StilwaterAI needs to publish well-researched content consistently.

The problem

Researching a topic properly, writing a first draft, and editing it to brand standards took about four hours per post. In an organization where every hour matters, that created real friction.

The StilwaterAI approach

We built what we preach. An AI-powered assistant called Blogbert researches a topic, synthesizes the findings, and writes a complete draft in the StilwaterAI voice, trained on examples of Jason Hurlbut's writing so the output sounds like him, not like a template.

The outcome

The whole process, from topic through research, first draft, and edit, now takes about 20 minutes. A task that consumed half a day is now a lunch break.