A single source of truth lets you use AI in your voice
This organization's content and supporter data lived in nine separate systems that didn't talk to each other — so nothing saw the whole picture, and off-the-shelf AI produced generic, off-brand output. I unified all nine into one organized, structured source, then built an AI content pipeline on top of it: it drafts in the organization's own voice, grounded in its own facts, and never publishes without a person's approval. The result is AI that sounds like them and works from their material — the opposite of generic.
AI customer support agent, running on your data
Customer email requests were arriving faster than the team could handle. I built an AI agent that works alongside the support team: for each conversation it decides how to pull the relevant customer information across four separate systems and provide a single view, ensuring the staff can respond quickly and accurately. They get draft email replies in the organization's voice, but the AI proposes and people decide — nothing goes out unreviewed. This turned a chaotic flow into a streamlined process.
Purchasing that always buys from the cheapest supplier
I built a complete product-catalog and ordering integration between a NetSuite ERP and three wholesale distributors. At order time the system compares each item's latest cost across all suppliers and automatically selects the lowest-cost source — providing significant annual savings.
Track every new supporter — and let AI surface who needs follow-up
Growing organizations lose track of who signed up, where they came from, and who needs follow up. I built a custom integration for NationBuilder that records how every new supporter entered the database, and tracks their activity and engagement over time. On top of this, I built AI assessments that find the 'needle in the haystack' — the specific people who are ready for personal follow-up, so organizers spend their time on the supporters who matter most.
A custom reorder-point system, built on your own sales history
Deciding when to reorder, and how much, is the critical middle ground between empty shelves and cash tied up in dead inventory. I built a custom reorder-point and demand engine from the business's own sales history — accounting for seasonal demand, supplier lead times, and the safety stock each product needs. This gave their purchasing team automated ordering guides for every one of 60,000+ products, across multiple locations.