Services / AI engineering
AI Engineering
I design and build agentic AI tools, chat functions, Claude skills and data products, from prototype to something your team runs every day.
What I build
I build AI software for specific jobs, not general chatbots. The best AI tools I have seen do one thing well: answer from a defined body of knowledge, run a repeatable research process, or turn messy sources into structured data a person can trust.
My flagship project is LandmAIn, the AI land intelligence platform I founded for data center and energy site selection. It runs from regional search to parcel-level analysis to a reviewable site brief, and every finding ties to a dated, inspectable source. That last part matters: an AI tool is only useful if people can see where its answers come from.
How I build
- Start with the evaluation. Before writing prompts, we agree on what a correct answer looks like and collect examples to test against.
- Keep humans in the loop wherever a wrong answer is costly.
- Prefer simple infrastructure. Static sites, serverless functions and plain files beat complex platforms for most internal tools, and they are cheaper to run.
- Ship in phases. A working prototype in weeks, then hardening, then features.
Tools for agencies and in-house teams
At Whole Whale I build agentic tools, Claude skills, plugins and browser extensions that turn an agency’s repeatable SEO, AEO and reporting work into one-click workflows. At Merge I built AI chat functions for hospital systems. The pattern is the same: find the process people repeat, write it down, and make it software.
Built for search, too
Because I also run answer engine optimization, the public-facing things I build come with clean markup, structured data and fast pages, so they can be found and cited from day one.
Related work
- LandmAIn (founder)An AI-powered land intelligence platform for data center, industrial, solar and storage site selection, with a dated, inspectable source behind every finding.
- Whole WhalePlugins, browser extensions, Claude skills, chat functions and reporting automations that package an agency's SEO and AEO processes into one-click workflows.
- Merge (enterprise healthcare clients)Conversational AI that helps patients find services and care information, built alongside AEO and technical SEO programs for enterprise health systems.
Questions
What kinds of AI applications do you build?
Mostly tools that sit on top of a team's own information: assistants that answer from your documents, agents that gather and summarize research, and data products that combine public sources into something decision-ready. I build with current models from Anthropic, OpenAI and Google and pick per task on cost, accuracy and data rules.
How long does it take to build an AI prototype?
A focused prototype that proves one workflow usually takes two to four weeks. A production tool with authentication, monitoring and documentation takes longer, depending on integrations. I scope in phases so you see a working version early and decide whether to keep going.
Who maintains the tool after launch?
You own the code and the accounts. I document how it works, keep the stack simple enough that a generalist developer can maintain it, and offer ongoing support if you want it. Hosting on Cloudflare keeps running costs low for most internal tools.
Further reading
Need help with AI engineering?
Email brubeckunhammer@gmail.com or use the project form. Tell me where things stand and what a good result looks like.