Add AI to my business
AI built into the systems you already run.
AI engineering here means building LLM features, MCP servers, and agents into the systems you already run, with human checkpoints where judgment matters, starting with a fixed-scope pilot that has a measurable outcome. It includes an honest scoping step: I will tell you when the answer is a script, not a model.
Also covers Automate my workflowsBuild an AI agent
- Fixed scope pilot, fixed price
- One workflow one measurable outcome
- Human checkpoints where judgment matters
- Mostly Claude swappable without a rebuild
Who this is for
- You have tried the chatbot demo and want something that actually runs your process.
- There is a repetitive workflow eating your team's time.
- You want AI agents that can operate your existing systems, not a separate tool.
- You want a straight answer on what AI can and cannot do for your business right now.
What you get
- LLM integrations built into your existing apps and workflows.
- MCP servers that let AI agents operate your systems safely.
- Automation pipelines with human checkpoints.
- Honest scoping.
Selected work
- CapCheck Multi-model cross-verification engine, automated claim extraction and normalization, RSS ingestion feeding LLM analysis.
- My own consulting platform Operated end to end by AI agents through an MCP server: client intake, statements of work, invoicing, and project data.
My take
I use AI the way an artist uses an assistant: it does the grunt work, and the human keeps every decision that matters. That is how this practice runs. My own consulting platform is operated by AI agents through an MCP server, but a person still approves every statement of work and every invoice before it goes out. The systems I build for clients work the same way, with a checkpoint wherever a wrong answer would cost something.
How it works
Workflow review
Where the time goes, and which parts a model can take over.
Pilot
One workflow, one measurable outcome, one price.
Expand or stop
Keep what worked. Drop what did not. No platform to commit to first.
The math
I've run an eight-person agency, held leadership in a hundred-person one, and been CTO of a third. I know where agency budgets go: overhead, junior hours, handoffs.
AI changed that math. The person who scopes the work is the person who builds it, and there is nobody in between.
Rate
Project
starting at
$3.5K
Scoped, built, and shipped to production.
Retainer
from
$500/month
Advise and build, ongoing.
Pilots are fixed scope and fixed price. Advisory is available as a retainer.
Questions
What is an MCP server and why would my business need one?
MCP, the Model Context Protocol, is a standard way to give an AI model tools: read this record, create that order, send this email. An MCP server is the piece that exposes your systems as those tools, with the permissions you choose. You need one when you want an agent to do work in your systems rather than just answer questions about them.
How do you keep AI from making decisions it should not make?
By deciding up front which actions need a person. Reading and drafting are usually safe to automate. Sending, paying, deleting, and anything customer-facing get a human checkpoint. The agent prepares the action and a person approves it, and every action is logged so you can see what happened and why.
What does a pilot cost and how long does it take?
A pilot is one workflow, scoped to a few weeks and priced before it starts. The workflow review picks the workflow and defines what a successful result looks like, so at the end there is a number to judge it by rather than a demo.
Which models do you use?
Mostly Claude, and whichever model fits a given task. The code is written so the model can be swapped without rebuilding the rest, because the models change faster than the systems around them.
Let's talk
One project at a time. Limited availability.
Or reach out directly:
Related
Last updated