Three named tiers. Clear deliverables. Everything in your name. Measured every week.
A productized AI consulting engagement for small and mid-sized businesses that want their website, search presence, and lead capture rebuilt on a modern AI-architected stack. Every AI tool runs self-hosted, on my hardware or on a Mac mini in your office, trained on the specifics of your business, tuned to convert your actual customers — not a template version of them.
The deliverable is on the page. Either the engagement works in month one or it doesn’t, and you don’t pay extra to find out.
Pick the scope, not a package matrix.
Foundation
For local service businesses with the basics in place but underperforming online. Get the technical foundation right so everything else compounds.
- Full SEO + technical audit (one-time, quarterly refresh)
- On-page SEO optimization + schema sitewide
- Google Business Profile overhaul + monthly posts
- Citation cleanup across 15+ directories
- 1 city/service landing page per month
- Monthly performance report
Growth — Most signed
Foundation plus conversion infrastructure and paid traffic optimization. This is where most businesses land — the technical base is built, paid traffic is dialed in, and a custom AI chatbot is capturing leads 24/7.
- Everything in Foundation, plus:
- Custom AI chatbot — self-hosted, your data stays local
- Google Ads audit + ongoing optimization (no commission on spend)
- Conversion tracking buildout (GA4, GTM, call tracking)
- 4 city/service landing pages per month
- 1 AI-assisted blog post per month
- Bi-weekly performance dashboards · Monthly 30-min strategy call
Dominate
The full stack. Built to make a business the dominant player in its local market. Currently signed at Dominate: Montgomery Water Softening (Conroe, TX) — active engagement since 4/20/26, with a performance dashboard delivered to ownership every week. In week one, on foundation work alone and before a single ad served, active users went from 15 to 72 and page views grew 8.8x. Where the account stands today, with the date each reading was taken, is in the case study.
- Everything in Growth, plus:
- Local Service Ads (LSA) setup + Google Guaranteed application
- 8+ city/service landing pages per month
- 2–3 AI-assisted blog posts per month
- Weekly performance dashboards · Monthly 60-min strategy call
- Priority response (4 business hours)
- Quarterly competitive intelligence brief
- Custom AI agent extensions (review automation, after-hours routing)
The posture, not just the deliverable.
Measured weekly
Every change has a written theory, and every report shows what worked and what didn’t.
Ad spend goes to Google
100% of what you spend on Google goes to Google. No middle layer taking a cut.
You own everything
Domains, ad accounts, GBP, content, code, model weights — all registered in your name from day one.
How every change ships
Every change ships with a written theory, a pre-declared abort condition, and a next-morning guardrail check. The abort threshold is re-derived before spend accrues, not after it. Experiments run one at a time; waste elimination ships immediately and gets logged.
Measurement runs unattended underneath all of it. A pipeline appends roughly 60 metrics a day to an append-only ledger, and when it misses a day the gap is written into the same ledger rather than papered over. A silent skip and a clean run look identical unless something is watching for the artifact.
We find the leaks nobody is looking for
On one account the phone-call conversion gate was set to 60 seconds, silently discarding 18 call-asset clicks every 30 days. Nothing was broken, nothing was alarming, and the account was quietly throwing away its own phone data. Reset to 30 seconds.
Automation with guardrails
Automation that fails loudly is safe. Automation that fails quietly destroys data. When the source sitemaps behind a dealer’s index-submission job returned nothing on 2026-07-29, the submitter aborted and left its state untouched rather than concluding the inventory had vanished and wiping its own known-URL list. A job that cannot tell “nothing new” from “nothing readable” will eventually delete the wrong thing.