Caleb W.

I put Claude to work inside the systems your business already runs. Most teams have AI open in a chat window while the real work still happens by hand in the CRM, the inbox and the spreadsheets. I connect the two: Claude reads your data, drafts the reply, sorts the request or writes the report, and you decide which steps a person signs off on. Recent Claude builds: • An SMS assistant for a martial arts school, built to answer parents' questions from the school's own knowledge base, now in a staff pilot • A daily German finance newsletter that Claude writes, with every edition approved by the client in Slack Other recent work: a lead pipeline that dedupes at intake so a duplicate is never written, and inbox triage that sorts ten categories and never answers a thread a person already answered. A lot of AI builds work in a demo and break on real data. Mine are built for real data: typed field mapping, deduplication on the way in, real failure states instead of silence, and a human checkpoint at every step you choose. My background is cybersecurity and data engineering. B.S. Cybersecurity and M.S. Data Science from Purdue, plus security internships at the Department of Defense and DNAnexus. What I build: • AI workflow integration: Claude connected through the API, with tool use and structured output • AI agents scoped to one task and grounded in your own data • Workflow automation in n8n, Make and Zapier • CRM automation across Airtable and GoHighLevel • Python services, API integrations and data pipelines • Web scrapers and document processing • No-code and low-code apps with auth, payments and data orchestration How I work: I start with the outcome you want, then design the simplest reliable system that gets there and ship it fast. Message me the process eating the most hours and I'll tell you where Claude fits in it.