Skip to content

The work

Systems running in production, not proofs of concept.

Client names are withheld. The systems are real, running today, and I'm happy to walk through any of them on a call.

Commercial fire protection contractor

214K files

A field-ops assistant that knows every job file

The problem

Years of contracts, specs, SOPs and job files sat on a server nobody could search. Crews called the office for everything, and QC follow-up depended on someone remembering to ask.

The system

An assistant in Slack trained on the company's 214,000 files and SOPs. It pulls project documents by job number, answers crews in English or Spanish, sends QC texts to the field and logs the replies back to the job.

Runs on
Slack, company file server, Monday.com, SMS
Replaced
Calls to the office, manual QC follow-up

Multi-clinic physical therapy group

Every clinic

Authorizations and visit caps that track themselves

The problem

Insurance authorizations and visit limits lived in spreadsheets. A missed expiration meant a patient turned away at the front desk or a denied claim weeks later.

The system

An operations system that stays in sync with the EMR, flags expiring authorizations and maxed visits before the appointment, and puts cases back on the schedule as soon as the paperwork arrives.

Runs on
EMR sync, internal dashboard, MCP for staff questions
Replaced
Authorization spreadsheets, front-desk guesswork

Defense-sector training company

Qualified only

A recruiting pipeline that qualifies before it books

The problem

Ads brought in applicants, but recruiters spent their day on calls with people who were never eligible.

The system

Google and Meta campaigns feeding an SMS flow that checks eligibility first. Qualified applicants get a booking link to a shared recruiter calendar. Everyone else gets a polite answer and no link. Qualified-lead data flows back to the ad platforms so they find more of the right people.

Runs on
Google Ads, Meta, LeadStorm AI, SMS
Replaced
Unscreened calls, manual applicant triage

Wealth management firm

One system

Outbound, content and replies on one system

The problem

Advisors were doing their own prospecting between client meetings, with no consistent follow-up and no content behind it.

The system

Cold email and LinkedIn outreach in each advisor's voice, articles published on a schedule, and replies routed straight to the right advisor so no warm lead waits.

Runs on
Cold email platform, LinkedIn, website CMS, reply routing
Replaced
Ad hoc prospecting, missed replies

Music school and AV production company

Page one

Local search at the scale of a hundred-plus pages

The problem

Two service businesses with real reputations, invisible in search outside their home city.

The system

Location and service pages built and published programmatically, over a hundred per site, with rank tracking and Search Console data driving what gets built next. The music school now ranks on page one for its core lessons in its target suburb.

Runs on
WordPress, DataForSEO, Search Console
Replaced
Agency blog posts that never ranked

Home inspection data platform (my own venture)

740K findings

98,000 inspection reports, structured and searchable

The problem

Inspection reports are PDFs. The data inside them, every defect in every house, was effectively locked up.

The system

An ingestion pipeline that turns reports from multiple inspection platforms into structured findings, removes duplicates, and serves them through a public API.

Runs on
Next.js, Postgres, Vercel, Claude
Replaced
Nothing. This data didn't exist in usable form.

Next step

Want one of these in your business?

Thirty minutes with me. You bring the business, I bring an honest read on where AI pays and where it doesn't.