TVLSS / Manufacturing
Primary practice · A

Manufacturing
& industrial.

Purpose-built software for Iowa plants that don't have a twenty-person IT team. We speak your stack — ERPs, PLCs, SCADA, spreadsheets, and that one Access database nobody wants to touch — and we ship modern tools on top without ripping anything out.

Book a plant walk → See the services
Why this practice exists

Built by an industrial engineer.

Most software vendors come at your plant from the outside. They learn your process by asking — and then they build a pretty UI on top of assumptions that don't quite fit.

I'm an industrial engineer. I came up in circuit-board manufacturing and fabrication — living inside SAP, QAD, and Paradigm (a process-based MRP we ran for both PCB and fab) — before I became a cloud and AI architect. When I look at your operation, I see takt time, changeovers, constraints, OEE, first-pass yield — the same things you do — and then I think about the software that makes those numbers move.

That's the difference. Same stack as the coastal AI studios. Different questions. Better software for your plant.

A.1–A.4 Services

Four tools. One floor.

Each of these can stand alone. Together, they turn a plant that runs on tribal knowledge into one that runs on data.

A.1
Serverless integrations

Move your data between systems that never talked.

AWS · event-driven

ERP to MES. MES to QMS. QMS to that shared drive finance uses. The plants that run best have pipelines that quietly move the right data between systems without humans re-keying it.

We build those pipelines as serverless functions — Lambda and SQS — so you're not babysitting a server, and you don't pay for idle. They scale to zero when the shift ends and light back up when the line restarts.

Legacy stuff doesn't scare us — ODBC, flat files on FTP, SOAP endpoints from 2008, CSV dumps from a nightly cron, proprietary XML formats from equipment built before Y2K. If it has an output, we can integrate it.

A.2
IoT & telemetry

Shop-floor data, useful for the first time.

MQTT · AWS IoT

You already have sensors. You already have PLCs with tag data. What you don't have is a place for it all to land — and someone to turn it into OEE, downtime pareto charts, and predictive alerts before the line actually goes down.

We meet the hardware where it is. ESP32 and microcontroller firmware, MQTT, industrial gateways. Data flows through Lambda into S3 or DynamoDB, then into dashboards your supervisors actually check.

Predictive maintenance is real, but it starts with boring work: clean data, tuned alarms, and a feedback loop with the people on the line.

A.3
Mobile apps

For the people who actually touch the line.

iOS · Android

Operator tablets that work with gloves on. Supervisor dashboards that survive a dropped phone. Field-tech apps that keep working when the WiFi doesn't.

We build offline-first. The app assumes the connection will drop. When it comes back, it syncs. Your people never lose a shift's worth of data because the router rebooted.

Distribution through Apple Business Manager and Google managed Play. We handle the MDM side and you handle running the plant.

A.4
AI copilots & vision

Tribal knowledge, finally searchable.

LLMs · RAG · vision

Your plant runs on SOPs, tribal knowledge, and the three people who remember why the line was set up that way in 2011. We turn that into AI copilots that your operators can talk to — "what's the torque spec on line 3?" — and get the right answer.

On the vision side: cameras over the line doing the boring QA tasks humans are bad at. Catching labels on backwards, seals off-center, and parts mixed up in the tote. Model training happens on your data, not a generic dataset.

Every AI thing we build is auditable. You see what it saw, why it flagged what it flagged, and what the confidence was. No black box.

§ Stack What we build on

Boring where it matters.

Proven, serverless, and chosen because it won't page you at 2 AM.

Cloud
  • AWS Lambda
  • API Gateway
  • DynamoDB
  • S3 · CloudFront
  • SES · SQS
IoT
  • ESP32 & microcontrollers
  • MQTT
  • Industrial gateways
  • Lambda pipelines
  • DynamoDB for telemetry
Mobile
  • React Native (cross-platform)
  • Swift · Kotlin (native)
  • PWAs when they fit
  • Offline-first SQLite
  • App Store · Play Store
AI
  • Claude (Anthropic)
  • GPT (OpenAI)
  • AWS Bedrock
  • Prompt caching

Walk us through
one thing that's broken.

Line downtime you can't trace. A report someone's building by hand every Monday. A copilot you wish your operators had. One problem is enough to start.

Book a plant walk → Send a message