Conference Presentation Deck

A Small Town That Refused to Wait

Little Schitt Creek Regional Library
How we built AI tools that respect privacy, serve patrons, and didn't cost a fortune

Who We Are

• Population: ~8,000
• Cardholders: 12,000 (regional)
• Staff: 8 FTE
• Location: Little Schitt Creek, Ontario
• Wi-Fi: Really good (we're proud of it)

We're not a big-city system with a tech department. We're a small library that got creative.

The Problem

• Search fatigue: 67% of catalog searches ended without a checkout
• Disconnected systems: 7 platforms, 7 logins, 23% digital resource usage
• Staff burnout: 12-minute average for "simple" questions, 60% of time on routine queries

What We Tried First

• Big vendor AI solutions: required cloud data transmission, cost more than our tech budget
• "Free" consumer AI tools: trained on everything patrons typed
• Doing nothing: patrons struggling, staff drowning

None of these felt right.

The Question We Asked

What if AI could run entirely on library property, using library values?

• No cloud
• No data leaving the building
• No training on patron queries
• No black boxes we couldn't explain to a skeptical board member

Our Six Principles

• Privacy first: session wiped when you close the tab
• Local control: all processing on library-owned hardware
• Radical transparency: every tool explains how it works
• Community input: patrons shaped the design
• Staff augmentation: helping librarians, not replacing them
• Bias awareness: tested for equity, surfaces diverse voices

The Digital Reference Desk

• Penny: conversational search assistant
• Resume Polish: grammar, action verbs
• The Jargon Buster: plain English for medical/legal forms
• Formal Emailer: professional message drafting
• AI Business Consultant: grant writing, business plans
• What Should I Read Next?: diverse recommendations by default

All running on a single Dell PowerEdge R750 in our server closet.

The Hardware

• Server: Dell PowerEdge R750
• Model: Llama-3-70b (hosted locally)
• Cost: ~$12,000 (one-time)
• Maintenance: shared across IT consortium

Less than one year of a typical vendor AI contract.

Community Co-Design

Before writing code, we held listening sessions. We brought coffee. We actually listened.

• Seniors: "Don't make me feel stupid"
• Parents: "Help me find books that look like my kids"
• Business owners: "I can't afford consultants"
• Teens: "Creative tools, not surveillance"
• Staff: "We're exhausted and honestly a little skeptical"

Meet Penny

Our AI assistant has a personality: warm, patient, like a neighbor who knows everything about the library.

• Seniors aren't intimidated
• Kids think she's fun
• Nobody feels judged for asking basic questions (there aren't any)

Personality is a feature, not frivolity.

Results: By the Numbers

Results: The Stories

"Penny helped me find books about my mother's village in Taiwan, books I didn't know existed. And she never makes me feel slow."
Eleanor Chen, 74

"My son asked for books with Black astronauts. The old catalog gave us nothing. The new tool gave us six."
Marcus Williams, parent

"The AI helped me write a grant. I got $15,000 for my food truck. The library didn't charge me anything."
Sarah Nguyen, business owner

Staff Impact

Before AI: 60% routine queries, 40% meaningful work

After AI: 25% routine queries, 75% meaningful work

Staff now do: 40+ new programs, community outreach, deep reference work, collection development that reflects the community.

Our Evaluation Scorecard

Every AI feature must pass ALL criteria:

Published on our website. Feel free to steal it.

Addressing Skepticism

• "This is a gimmick." Our circulation is up 34%. That's not a gimmick.
• "Small libraries can't do this." We're 12,000 cardholders with 8 FTE. We did it.
• "AI will replace librarians." We haven't cut a single position.
• "Privacy and AI are incompatible." Only if you let vendors define the terms.

Prefer a Human?

Technology is great, but sometimes you just need to talk to a person.

This isn't either/or. It's both/and.

AI handles the routine so humans can do the meaningful work.

Why This Matters

AI will reshape how communities find and use information. Libraries can either:

• Cede that future to companies whose values don't align with ours
• Shape it by building responsible models ourselves

We chose to shape it.

What You Can Take Home

• The scorecard: adapt our rubric for your own AI evaluation
• The approach: community co-design before code
• The permission: you don't need a big budget or tech team
• The proof: it works in a town most people only know from a TV show

Our Partner

We didn't do this alone.

The Unhinged Librarian (Sam Chada) helped with:

• Site design and development
• Staff training on AI tools
• Privacy-first implementation
• Community engagement strategy

unhingedlibrarian.com

Try It Yourself

We're happy to share: technical setup details, community engagement process, evaluation rubric, lessons learned (including the mistakes).

No spam, just librarians helping librarians.

Questions?

Thank You

Little Schitt Creek Regional Library

Empowering our community with knowledge, connection, and really good Wi-Fi.