A smarter site for caregivers.AI-first rebuild for families and caregivers navigating dementia care.
GLCF serves caregivers of individuals with dementia across the Greater Lansing area. Their old website was static: no search intelligence, no way to answer a question, no way to surface what was actually available locally. We rebuilt it from scratch with AI at the core: citation-aware answers, intelligent local search, and a platform caregivers can trust when the stakes are high.
The problem
GLCF serves caregivers of individuals with dementia, people navigating one of the hardest decisions of their lives. Their old website was static. It couldn't answer a question, surface local resources, or help a caregiver understand what was actually available to them in Lansing.
They needed something fundamentally different: a platform where AI understood the question, knew the local landscape, and responded with specifics. No generic search results or links to national aggregators. Something caregivers could trust because they could see exactly where every answer came from.
And it had to be sustainable. Local partners and sponsors were part of the ecosystem, but there was no structure that made those relationships valuable. No directory presence, no AI visibility, no tiered model that made a sponsorship worth buying.
What we shipped
We rebuilt the site from scratch, AI-first by design. Headless PayloadCMS for content management, Typesense for intelligent local search and RAG retrieval, Claude for conversational answers that cite every source. Caregivers search in plain language and get real answers, with the source shown, so nothing is taken on faith.
Local filtering keeps results Michigan-scoped by default, with a graceful fallback when local content doesn't cover the query. The AI searches GLCF's own partners and articles first, only reaching the open web when local knowledge runs out. No national aggregators crowding out community providers.
On top of the core platform, we layered a partner directory and tiered sponsorship model so the local relationships GLCF had built for years finally had a structure to make them visible and sustainable.
Built for the people who need answers most.
Every part of the platform serves one goal: give caregivers fast, accurate, local information they can act on.
Working on something similar?
Tell the AI what you're building and get a straight read on whether we're the right fit.
Three phases. One platform built around what caregivers actually need.
From a static old site to a full AI-first rebuild, then a partner intelligence layer on top.
Discovery & Architecture
Understand the caregiver. Design the stack.2025 · 2 weeks
Before building anything, we studied how caregivers of individuals with dementia actually search for help: what they ask, how they phrase it, what they trust. That shaped every architecture decision. Next.js for the frontend, PayloadCMS headless for content and partner management, Typesense for search and RAG retrieval, Claude for conversational answers. The stack was chosen to serve caregivers, not to be impressive.
“If the architecture does not serve the caregiver, we do not build it.”
Stack and scope reviewed and approved before any production code is written.
Build & Launch
AI at the core. Citations on every answer.2025 · 8 weeks
A complete rebuild of the GLCF website, designed from the ground up around AI-powered caregiver assistance. Typesense powers local-first search across GLCF content, partners, and resources. Claude handles conversational queries and returns answers with cited sources so caregivers can verify every response. PayloadCMS gives GLCF full control over content without touching code. Michigan-scoped filtering keeps results local by default, with a graceful fallback when local content runs thin.
Partner Intelligence
Local partners visible in every answer.2026 · ongoing
With the platform live, we layered in the partnership model. A partner-first RAG layer ensures GLCF sponsors and local resources surface before national aggregators in AI answers. External partner links became full on-site directory pages: measurable traffic, richer AI indexing, and a foundation for tiered sponsorship. Three tiers (Basic, Standard, Premium) give GLCF a product they can sell, with unmanaged listings showing a Claim Your Profile CTA that generates its own upgrade pipeline.
This project followed our four-phase methodology. See how every project runs →
Built With
Headless PayloadCMS for content and partner management, Typesense for AI-ranked search and RAG retrieval, Claude Haiku for conversational answers. GiveButter handles tiered membership payments; Mailchimp covers newsletter outreach. The stack is purpose-built for a community platform that needs to serve two audiences at once.
More work
View all
EdTech · classroom subscription
Lucky Little Learners
The problem: 30,000+ curriculum resources with no way to plan with them. Teachers kept falling back on the same materials.
Read the case study
Industrial printing distributor · AI-first e-commerce
PaladinID
The problem: Competing against manufacturers who sell direct, with a static website and a catalog buyers could not navigate.
Read the case study
Special education · membership product
The Intentional IEP
The problem: A library of IEP goals and six years of methodology that educators paid to access. The search returned noise. The platform was split across multiple logins.
Read the case study
See yourself in this story?
Most of our clients came to us with a similar situation. One conversation to find out if yours is a fit.
No commitment · No spec doc required.