// agents
AI agents
Tool-using agents that actually do work — research, outreach, triage, ops — with memory and guardrails so they stay on the rails.
Home/AI Development
// ai · agents · automation
AI Integration & Automation Developer
Custom chatbots grounded on your data, AI agents that do real work, Claude API integrations, and AI pipelines — built into your product or ops, not left as a clever demo. Done-for-you, by a developer who also teaches people to build agents.
// services
What I build
Grounded, useful AI wired into real workflows — and honest about when you don’t need it.
// chat
Chatbots & assistants
Grounded on your data (RAG), not the open web — support, sales, and internal Q&A that answers from your docs and cites them.
// integrate
Claude / LLM integration
Wiring the Anthropic Claude API into your existing product or workflow — cleanly, cost-aware, and without a rebuild.
// pipelines
AI pipelines
Classify, extract, summarize, qualify — batch or real-time, plugged straight into your stack.
// prompts
Prompt & content systems
Prompt engineering and generation tooling — the same discipline behind CookMyPrompt, applied to your use case.
// automate
AI-powered automation
The judgment-heavy automations rules can't handle — triage, routing, drafting, first-pass review.
// selected builds
AI I’ve actually shipped
Live products and real workflows — not a prompt in a text box.
CookMyPrompt
A live Claude-powered product
Turns a plain idea into a structured, high-quality prompt via the Anthropic Claude API — a real product with paying users, prompt-engineering baked into the pipeline, running on serverless functions.
Rivet
An AI companion, built into a product
The agent-style companion inside Skillwright — it reacts, upgrades, and guides learners through building their own AI agents. Designing AI that behaves well inside a real product is the whole job.
Grounded assistants
Chatbots & lead pipelines on your data
Claude-powered assistants and lead pipelines wired into real workflows — grounded on your docs and data (RAG), connected to email and your CRM, answering from what's true for you.
// build vs learn
Want it built, or want to learn?
This page is done-for-you — I build and integrate the AI into your product or ops. If you’d rather learn to build agents yourself, that’s exactly what Skillwright, my applied AI-agents course, is for. Same expertise, two ways to use it.
ClaudeAPI-first
RAGon your data
livein production
The stack
The Anthropic Claude API for the models, Python and Node on Netlify Functions for the plumbing, retrieval / RAG to ground answers on your data, Stripe and MariaDB where products need them, all version-controlled in Git. I keep AI features grounded, cost-aware, and boring where they should be — the same hand-coded discipline behind the software work and every product I ship.
// process
How it works
Scope & quote
We pin down the use case, where your data lives, and what a good answer looks like — plus a fixed scope and quote before any work.
Build & ground
I build it against your real data and test it on the hard cases, not a cherry-picked demo — with guardrails and a cost model that makes sense.
Ship & support
Deployed into your product or ops, documented, and supported after handoff — with your keys and your data staying yours.
// questions
FAQ
Chatbots and assistants grounded on your data, AI agents, Anthropic Claude API integrations, and AI pipelines (classify, extract, summarize, qualify) — all wired into real workflows, not demos.
Yes — grounded on your data with RAG, so answers come from your docs rather than the open web. Support, sales, and internal Q&A that cites its sources.
An agent is an LLM that can use tools and take steps toward a goal, not just chat. If you've got a multi-step, judgment-heavy task, an agent can run it; if you just need answers, a chatbot is simpler and cheaper. I'll tell you which you actually need.
Primarily the Anthropic Claude API, picking the right model for the job so you're not overpaying for simple tasks. I can work with other providers where it makes sense.
It stays grounded on your data and your API keys. I don't train on it or route it anywhere it doesn't need to go, and we scope data handling before anything is built.
Yes — Skillwright is my applied course on building AI agents. This page is the done-for-you version; Skillwright is the learn-it-yourself version.
// free knowledge
From the guides
The write-ups behind the services — free, no gate.
// get in touch
Ready to put AI to work?
Tell me the workflow you want smarter — you’ll get a straight answer on whether AI actually helps, a scope, and a quote.