Need an AI agent?
Let's build it right.
I build production-grade AI agents and automations for your team — from scratch, on your stack, on a per-project basis. Already have an agent that half-works? I also unstick the ones that stall between demo and production.
Four ways I can help
Hire me per project. We scope the work up front, so you know exactly what you're getting before anything starts.
Build a new agent
No agent yet? I build production-grade agentic systems from scratch: custom tool-calling frameworks, RAG over your data, and the integrations to put it where your team already works. Or get a scope for your potential project to help an internal team get started.
- Custom Python tool-calling frameworks
- RAG with vector DB & embeddings
- Integrations (Teams, Slack, Excel, web)
- Document/OCR pipelines
Ship it to production
The last mile is the hardest. I make an agent reliable, observable, and deployable, with the guardrails, logging, and hosting it needs to run in the real world.
- Logging, tracing & observability
- Hosting & deployment (local LLM or cloud)
- HIPAA / regulated-environment experience
Unstick a stuck agent
You have an agent that half-works. I dig into the prompts, tools, retrieval, and control flow, find why it breaks, and fix the root cause, not just the symptom.
- Tool-calling & control-flow debugging
- Retrieval / RAG quality fixes
- Grounding to prevent hallucinations
- Evaluations for measurable progress
Support the tool you already built
Did your team develop an incredible internal tool? Maybe used AI coding agents to develop it, but now can't spare the manpower to support it in the cloud? We can help. Contact us for more information.
- Cloud hosting & deployment
- Monitoring, logging & alerting
- Dependency, model & security updates
- Ongoing fixes and small enhancements
Upskill your team with live AI training.
Want to upskill your team? Build your own automations? I run hands-on, in-person or virtual live sessions where we build real agents together on your stack. Your team can start building right away, the right way.
Agents I've built and can build for you
I'm Noah McGehee, an AI engineer with a systems-engineering background. I'm experienced in working and deploying AI workloads in HIPAA compliant cloud environments. Here's a list of few agents and products I've built for other regulated businesses:
Knowledge & document assistants
An agent that answers questions from your own documents, wikis and data, with citations and per-user permissions. The retrieval is tuned to your content, so answers are relevant in depth and scope.
Ops & monitoring agents
An agent that watches your systems, triages, and details problems with the evidence attached. Read-only access and deterministic retrieval stop it from inventing a diagnosis or editing a live environment
Workflow & integration agents
Agents that live where your team already works, whether that is chat, a spreadsheet or your own web interface, with custom tool-calling so they take real actions instead of just describing them.
Document & intake pipelines
Agents that read PDFs, scans and spreadsheets, pull out what matters, and hand back structured data or a finished document.
Regulated & self-hosted builds
Agents for places where the data cannot leave: local or on-prem models with access controls per user. Or cloud deployments with compliant data residency.
The stack underneath
Local LLMs, small models or frontier APIs, chosen for your privacy level, latency and budget constraints, plus the retrieval, evals and infrastructure that keep things reliable.
Let's build your next agent.
Book a short call — tell me what you want built, or where an existing agent is stuck.