The Pilot Library

The Open Playbook

Every pilot we complete contributes to the Applied AI Playbook — an open-source, permissively licensed library of rural-contextualized use cases, implementation guides, and evaluation frameworks published for any rural organization to adapt and deploy.

The Center for Rural AI is itself becoming an AI-native organization. We are building how we operate daily for research, outreach, and program delivery around AI from the ground up, so that our guidance to rural institutions comes from lived experience rather than theory.

Running this way means we feel the real challenges and the real benefits of adopting AI firsthand: the workflows that break, the ones that transform, and everything in between. That hard-won understanding is what we bring to the organizations we serve. The agents below started as tools we needed ourselves — then became pilots other rural organizations can install.

Our aim is for our AI-native design to add capabilities a small team wouldn't otherwise have access to, the equivalent of a marketing lead or a dedicated technical-support person and to roughly triple the effective output of the staff we do have.

Why It Matters

Because we operate lean and AI-native, we can show what a small rural organization can accomplish with the right tools and pass those playbooks directly to the institutions and communities we support.

For a small nonprofit, finding the right grants is a job unto itself, one few teams have the hours to do well. The Grant Writing Assistant scans for grant opportunities, assesses how well each one fits the organization's mission and eligibility, and compiles the qualified ones into a living database. Every entry carries the details that matter for a go/no-go decision: deadlines, award size, and requirements.

What makes it reusable is the work underneath. We turned CRAI's own Organizational Reference Document into a fill-in template that any organization can complete to get started. Feed the completed template to the startup skill and it builds the whole system for you, the Airtable database, the scanning and scoring skills, and the audit trail. It's a skill that builds skills.

The Org Seed Template

The single most important artifact is the Org Seed Template: 41 questions across 13 sections that build a comprehensive, structured view of an organization. It moves an organization's canonical information out of a loose document and into a fill-in-the-blank template that becomes the source of truth for everything downstream.

Reusable AI Skills

Four reusable AI skills built so far, the building blocks another rural organization can adapt.

Proven End to End

To prove the model, we built a fictional nonprofit deliberately unlike CRAI and ran it through the whole pipeline: the agent took its completed template, generated the Airtable base, populated every table, and left it ready for its first scan. In testing, the onboarding skill built every skill and Airtable asset in about five minutes. Early onboardings can be walked through in a short session covering account setup, the skill install, and a first guided scan.

The assistant runs in production at CRAI, and the first install at an outside organization was completed in July 2026. The public repo carries the full install runbook — the validated production setup plus the failure modes observed during live installs and how to avoid them — so the next organization starts from a proven path rather than a blank page.

CRAI has developed a replicable AI pilot model for independent rural inns, B&Bs, and lodges. The pilot focuses on practical, low-cost workflows that help small hospitality operators recover time, enhance their SEO, improve guest communication, identify new market segments, and strengthen shoulder-season revenue.

Rural hospitality businesses live or die by bookings, but the work of filling a calendar — finding the right event planners, tracking down real contact details, writing outreach that connects, and getting found on search is exactly the kind of time-consuming task a small owner-operator never has time for. This pipeline handles the hardest parts end to end: it locates the right firms, tracks down verified contacts, drafts outreach that sounds like it came from the owner, and produces search-optimized content in the owner's own voice that lifts the property's visibility without sounding like a marketing department.

It was built for one rural inn but designed from the start to be reused by the next. The pilot is active with its first partner site, so outcomes are still in progress, and each workflow is being documented as it proves out.

Reusable AI Skills

Four reusable AI skills built so far — the building blocks another rural organization can adapt.

Proven End to End

The pilot's proving ground is Gable House, an independent inn in Durango, Colorado. Working alongside a real owner-operator keeps every workflow grounded in the day-to-day realities of running a small rural property — and produces the documentation the open playbook is built from. Five target workflows are in scope:

• ▪AI-assisted market segmentation — identifying new guest segments worth pursuing.
• ▪Group booking outreach — finding and reaching event planners and other group organizers.
• ▪SEO content in the owner's voice — publishing keyword-rich blogs and page copy that raise the property's organic search ranking while sounding genuinely like the owner.
• ▪Pre-arrival personalization — tailoring guest communication ahead of each stay.
• ▪Stay-extension prompts — surfacing timely offers that lengthen bookings and support shoulder-season revenue.

“We're still in the middle of building an AI hospitality pilot, but I can already tell it's going to positively impact our brand and bookings. What I appreciate most is that whatever we figure out together, other small hospitality businesses like ours will get to use it too.”

Open-Source & Configurable

The intended public output is an open playbook — the prompts, workflow documentation, and low-cost tool templates behind the pilot — so other rural inns, B&Bs, and lodges can adapt the same workflows to their own market without starting from scratch.

A team member posts the natural language change they want in Slack, an automated Claude task drafts the update, and a maintainer confirms it before it goes live.

The key is the human approval step: nothing touches the live site until a person reviews and signs off, so the convenience of plain-English requests never comes at the cost of control.