AI-native product studio

From the dataset to the device.

I build AI-native software systems end to end — the owned datasets, the pipelines that keep them fresh, the platforms on top, and the native apps that ride them. Deep in freight and logistics; the pattern travels.

2 product systems 5 stack layers MCP in every product

The portfolio

Two systems, owned end to end.

Most shops build the app and rent the data. I build the whole column — the dataset, the pipeline that keeps it fresh, the platform on top, and the native app that rides it. Here are two, running today.

Sys · 01

Freight Intelligence

Find the shippers, vet the carriers, move the loads — one stack from raw records to a running TMS.

Data Fast Crawls · FMCSA · Enrichment — 30M businesses + national carrier records, AI-scored
Platform FreightBrokerLeads.ai · Freight Wranglers TMS — spatial lead-gen + full dispatch & billing
Surface Web · REST API · MCP · mobile — an AI can search leads and run the board
Broker leads · carrier vetting
demo.freightwranglers.com/dispatch/board
Freight Wranglers dispatch board — a live Kanban of loads by pipeline stage
freightbrokerleads.ai
FreightBrokerLeads.ai — shipper density map and lead stats

Sys · 02

Truck Navigation

Truck-legal routing that dodges low bridges and works with no signal — on a continental map I own, from Chiapas to Alaska.

Data OSM · Overpasses · TruckNavPipeline — AI-verified clearance for ~1M overpasses + restriction data
Engine MapTilesValhalla — I build the routing tiles, offline maps, and geocoder
Device Truck Magnet app + platform — on-device Valhalla, fully offline · iOS · Android
Copilot Hands-free voice assistant — find + add stops, ask for info, plan the trip while you drive
251M address points More than OSM or OpenAddresses Routes fully offline
iOS + Android · private beta
On-device · offline
Truck Magnet turn-by-turn navigation — a truck-legal route with speed, ETA, remaining miles, and live hours-of-service
Truck Magnet HOS-aware trip planner — arrival time, drive time, distance, and on-time status
Truck Magnet · Android

More from the studio

The pattern travels.

Two more, same DNA — a field-service platform that proves the model works beyond freight, and the AI control plane I run all of it from. Every one ships with an MCP server, so an AI can operate it.

Field-service platform

MyBizEZ

The whole stack for home-services contractors — quoting, scheduling, crews, billing — on web and native mobile, with an AI that drafts quotes from a spoken note.

new quote — 3 oak removals, haul the debris
draft quote created from voice · ready to price
Web + native mobile6 payment gateways Voice-to-quote~2,750 tests

AI infrastructure ops

MCG.Work

The control plane I run everything from. One MCP server: talk to it and it edits DNS, issues TLS certs, fixes email, and reconfigures a live firewall.

stand up scalehouserocks.com
cert issued · HAProxy backend on 3 nodes · DKIM set
DNSTLS certsEmail / DKIM HAProxypfSense firewall

Approach

AI-native, all the way through.

The AI isn't a feature bolted on at the end. It's in the product, the interface, the build, and the stack — engineered for your operation and tested until it holds up in the real world.

In the product

AI that fits the operation

Not a chatbot in the corner — AI built into the real workflow. A crew finishing a job taps once and talks; the estimate writes itself.

add maple removal $1,400, stump grind $800, haul-away $1,500
3 line items added · estimate updated · $3,700 total

Right model · cost + token analysis · tested in production

In the interface

An MCP server in every product

Your software becomes something an AI can operate — a copilot for your team and your customers, in Claude, ChatGPT, or your own app.

Conversational access · scoped · rate-limited

In the build

The build itself runs on AI

I don't hand-write code line by line — I direct it, often by voice, and AI writes it. Then AI reviews every line that ships for correctness, security, and best practices — faster, and more thorough than hand-review ever was.

Voice-directed · AI-written · AI-reviewed

In the stack

Owned data, self-hosted from the metal up

The datasets, the pipelines, the routing and geocoding, the firewall and mail — all mine to run. No per-seat data rents, no vendor lock-in.

Data · pipelines · routing · firewall · mail

Hardened

AI you can trust in production

Every model integration ships with guardrails, prompt and model testing, and injection defense — so the AI does its job and nothing else, even when someone tries to make it misbehave.

  • Prompt guardrails
  • Model & prompt testing
  • Prompt-injection defense

Who’s behind it

I’ve hauled the freight and built the software.

I’m Michael Gehring — a developer with 30+ years in software and a CDL-A who ran 49 states hauling arms, ammunition, and explosives for the DoD before spending the last year building AI into operations management systems. I know freight from the cab and from the codebase.

AI, software, and computers are what I do for fun — I stay on the leading edge and use three decades of experience to tell the durable from the hype. Husband and father of three; I lead by example and ship things that work.

30+ years in softwareCDL-ASecret ClearanceTWIC AWS CertifiedYoder, Indiana

Contact

Open to the right role — or the right build.

I’m looking for a place to build AI-native software for a company that moves things — and I take on select client work too. Freight and logistics especially. If that’s you, let’s talk.

An unhandled error has occurred. Reload 🗙

Rejoining the server...

Rejoin failed... trying again in seconds.

Failed to rejoin.
Please retry or reload the page.

The session has been paused by the server.

Failed to resume the session.
Please retry or reload the page.