Hand-drawn cartoon of Dee on a stool, arms folded, watching two robots chat on a monitor.

One AI integration, scoped and shipped in a week.

I find the spot where AI saves you real hours, build it, document it, and hand it off so your team runs it without me.

$3,000 flat 1 week

Pick the one that fits.

For your business

Plugged into your product or your team's process. Saves real hours on the work a person was doing by hand.

  • Internal tool. Custom UI on top of Claude or OpenAI, scoped to one team's job.
  • API pipeline. Classifying, cleaning, or routing data between systems you already run.
  • Chatbot. On your site or in-product, answering from your own docs.
  • Workflow automation. n8n / Make / Zapier glued to an LLM, replacing the manual step.

For yourself

An agent running on your hardware, wired into your files, apps, and shortcuts. Private, always on, yours.

  • Local agent. Mac or Linux, into your shortcuts, files, and apps.
  • Headless on a VPS. Runs 24/7, picks up tasks, reports back.
  • Custom MCP setup. Your data sources, your tools, your model of choice.
  • Browser / computer use. For the tasks that still need clicking around.
  1. Pick one workflow or opportunity

    The $3K build is for one scoped integration. If you're not sure what should be automated, start with the $500 audit: I look at one workflow, team process, or personal setup and tell you where AI actually belongs.

  2. The honest filter

    If a forty-line script solves your problem, I won't sell you a language model. The first thing I do is figure out whether AI is the right tool. Sometimes the answer is no, and that's worth knowing on day one.

  3. How the week shapes up

    Day 1 the brief lands. Day 2: architecture sketch and a working spike on staging. Mid-week: production build, real edge cases, real data. Friday: docs, walkthrough, monitoring. Handed off so you run it without me.

  4. Where your data lives

    Personal-agent setups run on your hardware. Workflow integrations depend on the provider, and I'm explicit about where data flows before we start. Repo in your GitHub from day one. Full refund at the 48-hour checkpoint if you walk.

Tell me the workflow. I'll tell you if AI helps.

Not sure where AI belongs? Start with the audit: one workflow, one opportunity map. The full $500 rolls in as credit.

A few things people ask

Should I start with the audit?

If you already know the exact workflow, no. If you're asking 'where could AI help us?' or 'what can I automate?', yes. Audit one workflow or setup first, and the $500 rolls in as credit if we build it.

What models / providers do you use?

Whatever fits the job. Claude, OpenAI, local Llama via Ollama, open-source embeddings. I pick based on the constraint, not the brand.

Do you do RAG / multi-agent / fine-tuning?

RAG and multi-agent yes, often. Fine-tuning rarely. Most problems are better solved with a sharper prompt and a smarter retrieval layer.

Other things I do

  1. Audit One focused thing. Credited if we build. $500
  2. Landing page Deployed on your domain. $3,000
  3. Idea to MVP Working product on your domain. $9,000
  4. AI visibility Get found by AI search. $3,000
  5. Anything else Custom scope, custom shape. Custom
Send a brief