# The Mad Botter INC — full reference

> The Mad Botter builds operational data infrastructure for environments where the cloud is not an available answer — capturing the record on the equipment where work happens, inside your compliance boundary, and proving it by replay.

Positioning, in one line: operational data infrastructure for environments where the cloud is not an available answer.

The moat: The Mad Botter owns the moment operational truth is created — on physical equipment, inside a compliance boundary — and makes that record provable afterward. Three conditions make that defensible, and every page on the site touches at least one. The data is born on physical equipment: sensors, serial devices, printers, scanners, PLC-adjacent hardware, captured with kernel-level timestamping and store-and-forward through outages — you cannot prompt your way to that. It is born inside a boundary: ITAR/EAR, CMMC and NIST 800-171 posture, GxP, air-gapped and expeditionary sites, facilities where data cannot leave the building, where cloud tools cannot enter the room at all. And the record must be provable: deterministic capture at the point of work, a hash-chained append-only ledger, and replay verification that anyone can run independently on a laptop.

The three published acceptance criteria, stated as falsifiable tests rather than adjectives: zero record loss through a comms outage, every record re-verifiable against the ledger, and the whole chain demonstrably verifiable air-gapped on a laptop.

The problem being replaced: on most lines operational truth is reconstructed after the fact, from memory, paper, and spreadsheets. A reconstructed record fails the same three ways — it fails the audit, it fails the dispute, and it fails the AI initiative that assumed the data existed.

Writing code got cheap. Operating production systems did not. Anyone can generate something that works in a weekend; almost nobody can take a twenty-year-old schema nobody can explain, a packing line that talks over serial, and an export-control boundary, and be accountable for the result on Monday morning. That knowledge was never written down, so it was never in a training set. The scarce thing moved from code to accountability, and accountability is the product.

What that means concretely: undocumented legacy systems, hardware and edge integration, regulated and air-gapped deployment, and self-hosted infrastructure The Mad Botter operates rather than hands off. Not prototypes — the thing that has to be running.

The public front door is the Equipment Fit Check at https://themadbotter.com/fit-check: free, no payment, no meeting required, and a written one-page verdict within two business days on whether the platform captures the customer's equipment out of the box. It is not an audit, a quote, or a discovery workshop, and the verdict is the customer's to keep whether or not they ever buy anything.

Two funnels sit behind that door and they share no price, page, or next step. Equipment and physical operations: Fit Check (free) then First Line Pilot ($1,500 flat) then full platform deployment. Recurring digital workflows between systems: the Workflow Proof — $1,500 fixed, one recurring manual workflow, a working proof against real or representative data in five business days, fit checked before payment. Alice is the reusable governed workflow engine behind the work — connectors, transformations, approvals, run history, scheduling, and output. Vorpal is the physical-operation-capture layer, for cases where the missing record lives in equipment rather than software; it is scoped separately and is not the primary conversion path. Neither is a prerequisite to buying the Workflow Proof, and a buyer does not have to learn either name before the first purchase.

The company is small, senior, and 100% US-based. The Mad Botter is designed to support a client's compliance posture (CMMC 2.0 / NIST 800-171, ITAR, EAR, CUI); it does not certify compliance.

## What The Mad Botter is not

- Not a generic "we build anything" development shop.
- Not an AI-readiness consultancy.
- Not a collection of unrelated SaaS experiments.
- Not a lead-generation company.
- Not a prediction or forecasting product. Capture records what happened; it does not guess what will happen.
- Not an ERP replacement in every engagement.
- Not a staffing or developer-capacity vendor.

AI, IoT, dashboards, and automation are capabilities. They are not the category-level pitch.

## Who it's for

Operational teams carrying recurring manual work between systems, plus the regulated, asset-heavy operations that are the deepest proof of the pattern — aerospace and MRO, AEC/heavy civil, pharma/CDMO, and industrial agriculture. Recognizable symptoms:

- "Only one person knows how to build it" — a report, reconciliation, or handoff that lives in one person's head.
- The numbers change depending on who exports them, or the report is late whenever the expert is out.
- The workflow crosses systems nobody wants to replace, so the recurring cleanup happens by hand every cycle.
- The ERP, maintenance system, project platform, quality system, or BI tool records the transaction but not the workflow around it.
- Compliance evidence must be assembled manually before an audit, without touching validated production systems.
- A previous custom system has become difficult to maintain, or "just put it in the cloud" is not an available answer — ITAR, EAR, CMMC, NIST, or an air gap.

Operational buyers: owner or president, COO or VP Operations, director of operations, director of maintenance, warehouse or production leader, program or project executive. Technical and risk buyers: CIO, CTO, IT director, IT/OT leader, compliance or quality leader, security or infrastructure owner. The site also serves referral partners — ERP and MRO consultants, MSPs and fractional CIOs, aviation and defense consultants, compliance specialists, prime contractors and systems integrators, equipment and telemetry vendors, and private-equity operating partners.

## What an engagement includes

Fixed-scope operational software, not a services menu to choose from:

- The Workflow Proof — one recurring manual workflow, real or representative data, a working Alice proof in five business days, an agreed output, a run record with assumptions and exceptions, a walkthrough, and a fixed production quote.
- Alice — the governed workflow engine. Connectors, transformations, approvals, run history, scheduling, and delivery for the recurring digital work between systems.
- Vorpal — physical-operation capture. Gateways, scales, label printers, barcode scanners, kiosks, and serial-line equipment, for when the missing record lives in equipment rather than software. Scoped separately from the Workflow Proof.
- Production implementation and managed operation — a fixed quote before work begins, then monitoring, upgrades, and incident response contracted against an agreed service level under a separate operations agreement.

The customer gets: one accountable US-based engineering partner; a known price and a known deadline before the first line of work; software adapted around the actual business rather than a generic template; migration from spreadsheets, databases, documents, and legacy systems; integration with systems that must remain; self-hosted deployment and control of operational data; and ongoing support, extension, and stewardship after production.

## Offerings & pricing

- **The Equipment Fit Check — free.** The public front door and the site's single primary call to action, at https://themadbotter.com/fit-check. The customer lists equipment and environment — protocols, makes and models, how many lines or asset classes, and any air-gap, ITAR/EAR, CMMC/NIST 800-171, or GxP constraints — and a senior engineer returns a written one-page verdict within two business days, reviewed against the supported-equipment matrix rather than generated. Every item comes back marked supported (standard gateways and drivers), needs a scoped driver (capturable, and priced before the engineering begins), or out of scope (said plainly, with a pointer elsewhere where one exists). The verdict also states how the flagged environment constraints are handled, and — if the customer runs the pilot — which line would be started with and the machine-verifiable acceptance criteria that would be proposed. There is no payment step, no meeting is required to receive it, and there is no follow-up sequence beyond the verdict itself; a green verdict carries a link to schedule the First Line Pilot call. Explicitly NOT an audit, a quote, a commitment, or a discovery workshop.
- **The Workflow Proof — $1,500 fixed.** The first step in the other funnel — recurring digital workflows rather than equipment on a line. One recurring manual workflow, the customer's real or representative data, a working proof in five business days. Includes one representative input set, one agreed output, one review cycle, a working Alice proof, a run record, a walkthrough, and a fixed production quote. Excludes live production deployment, changes to validated systems, new hardware or device integration, multiple workflows, unlimited cleanup, ongoing support, and open-ended time-and-materials. Fit checked before payment: the intake form at https://themadbotter.com/start takes no payment — it emails a request, Michael replies within one business day with a fit decision, and a written scope note (input, output, acceptance test, exclusions, timing) follows if it fits. The full $1,500 credits toward a fixed-price production implementation if the customer continues. Production implementation is a fixed quote first: simple live deployments typically from $3,500, and multi-system, on-premises, or regulated implementations typically from $7,500. Managed operation is quoted per engagement and carries no published rate.

- First Line Pilot — $1,500 flat, an unrelated offer at https://themadbotter.com/first-line-pilot that happens to share the figure with the Workflow Proof. One production line, thirty days, verified by replay at day thirty. A preconfigured Core appliance and up to three Capture gateways — loaned from The Mad Botter's own pool, not sold — ship to the customer's dock; their own maintenance tech or electrician mounts them from the included guide, and bring-up is a scheduled remote session — no contractor and no site visit. The pilot is requested, not bought on the spot: a short intake form that takes no payment and collects no card details. Michael reads the request, confirms the line is one we can start on, and raises an invoice by hand — ACH, wire, or against a purchase order on net terms. There is no sales call and no assessment, and submitting the form commits the customer to nothing. The qualification criteria are published on the pilot page so the customer can assess their own line before asking — a US or Canadian facility, one production line or asset class whose equipment can be physically reached, and power within reach of that equipment — and we check the same three before invoicing. An optional twenty-minute call can be booked from the pilot page before ordering — no pitch, same price — but nothing requires it. Every pilot is invoiced; where procurement needs a PO number and billing contact captured up front, those are supplied at https://themadbotter.com/first-line-pilot/invoice instead of by reply. The buyer protection is both the refund and the fact that a human decides whether to invoice at all: cancel any time before the kit ships for a full refund, no reason required, and after it ships one trigger remains — if bring-up cannot produce events from the equipment the customer described when ordering, that is a full refund too and the kit returns on the prepaid label. The box ships within five business days of a confirmed line: after the invoice is raised the buyer is asked by email what is on the line and where it goes — makes, models, counts — because the kit is built to that list, and one reply starts the clock. Bring-up is booked within fourteen days of delivery; the thirty-day clock starts at the first captured event, or fourteen days after delivery, whichever comes first. The architecture write-up delivered at day thirty carries the hardware specification, the deployment sequence, and the price of the rollout it describes. The full $1,500 credits against the first platform invoice if they continue. The loaned kit returns on a prepaid label either way — it is never sold. Production runs on commodity units the customer buys themselves, to a specification we provide. Escorted on-site install is $2,500 flat. Explicitly NOT a paid assessment and it does not ship a report — a prior $5,000 five-day assessment offer was retired because selling a document is a different business. Not linked from the homepage or /start; reachable from the footer only.

- Self-serve pipelines pilot — $149/mo (Workspace tier) with a 14-day trial. One source, one Power BI dataset, a scheduled refresh, live before the next reporting meeting. Connect, inspect what the platform proposes, approve. No sales call. Linked from /platform only, not from the Workflow Proof funnel.
- Full platform deployment (First Line Pilot funnel only) — $1,500 per line per month after a pilot. Hardware is separate and bought outright by the customer from their own supplier; The Mad Botter does not resell equipment, so no hardware figure is published. Scope is still per engagement — how many lines, what has to be integrated, what the compliance boundary demands — and the pilot's day-thirty architecture write-up specifies the units and the timeline, so the customer can check the number before committing to it. The $1,500 pilot credits in full against the first invoice. Unrelated to the Workflow Proof's own production-implementation guidance above.
- Forward-deployed engineering — bounded ownership of a defined part of the production stack rather than billed hours, delivered through a documented playbook with a named lead. Managed operations run against an agreed service level under a separate operations agreement.
- Migration and integration — moving legacy data and connecting the systems that remain, without a rip-and-replace cutover.

## How an engagement runs

The Workflow Proof, day by day:

1. Day 0 — fit and finish line. The workflow is reviewed, confirmed to fit, and the input, output, and acceptance test are documented in writing. Payment follows acceptance of that scope, not the other way around.
2. Day 1 — data and rules. The customer provides a representative export and explains the current decisions, corrections, and exceptions.
3. Days 2-4 — build and test. The workflow is implemented, results are compared against the current process, exceptions are exposed, and the repeatable run is prepared.
4. Day 5 — replay and review. The agreed input is run, the output is produced, the result is walked through, and the fixed-price production recommendation is delivered.

The five-business-day clock starts once the usable input and the written acceptance test are confirmed — not from the moment the intake form is submitted. What happens after the form: a fit decision within one business day; a written scope note (input, output, acceptance test, exclusions, timing) if it fits; then, once the proof is approved and paid for, a secure data-transfer path and the day-five review.

After the proof, the customer decides on production. Continuing within the approved credit window applies the full $1,500 to a fixed-price production implementation — a separate, larger engagement with its own scope note, live sources, deployment, scheduling, and acceptance test. Nothing converts automatically into open-ended time-and-materials.

The customer keeps their source data, outputs, customer-specific configuration, documentation, and deployment artifacts. The Mad Botter retains the reusable Alice and Vorpal core technology and licenses it.

What does not fit the Workflow Proof, stated plainly on the site: a request whose finish line cannot be defined; several unrelated workflows bundled into one ask; live writes into production or validated GxP systems; hardware, PLC, scanner, sensor, or serial-device integration (that is a Vorpal conversation, scoped separately); a new custom application or user interface; on-site work; and undefined cleanup of an entire data estate. Those can still be good Mad Botter engagements — they are quoted separately rather than squeezed into the $1,500 box.

## Deployment and control

Self-hosted on infrastructure the customer owns — bare metal, their own servers, or a private cloud tenant. Container-based delivery on PostgreSQL and established production technology, so an existing IT team can manage it with familiar tools. Customer-owned infrastructure and operational data, least-privilege access, audit logging, integration with the systems that must remain, and air-gap-capable architecture where required. Technical detail is there to reduce perceived risk, not to be the value proposition.

## Proof

### Aerospace and MRO — a DC-area aerospace parts broker and MRO shop (2026)

The customer supplies and overhauls components for defense and government customers, under MIL-STD-129 marking rules and ITAR/EAR export controls, with high-value corrosion-sensitive inventory.

Before: day-to-day work spread across fifteen Monday.com boards holding roughly 14,000 line items, a tangle of Google Sheets, a 1.6-terabyte shared document store, QuickBooks, and hand-built Word and PDF templates. An order's history physically migrated from board to board as it aged — RFQ, orders, warehouse, finance, archive — stitched together only by free-text reference numbers typed by hand. RFID tags were peeled off a spool and scanned into the DoD tracking system with nothing in the company's own records linking tag to job. MIL-STD-129 labels were produced by screenshotting a third-party tool and eyeballing the print scale. Stored components had zero temperature or humidity monitoring.

Built: a customer-specific extension of the platform as the single system of record; ~14,000 records migrated out of the fifteen boards; field gateways (BeagleBone Black and Raspberry Pi over ethernet, Wi-Fi, or cellular) authenticating with rotating BCrypt-hashed keys and streaming temperature, humidity, pressure, vibration, and moisture into a time-series store; a rule engine with per-asset and per-equipment-type thresholds raising graded alerts to email and Slack; RFID records cross-referencing job, MIL label, Transportation Control Number, and DoD scan; MIL-STD-129 labels streaming to Zebra printers as ZPL with Code 128 and Data Matrix/PDF417 barcodes; barcode cycle counts against printed bin tags; and structured receiving with a four-stage repair-photo record.

Results: a live operational picture instead of a rear-view mirror. The single most valuable win is environmental monitoring of stored inventory — humidity and temperature alerts now warn of corrosion risk before high-value aerospace components are damaged. Chain of custody is unbroken, with every move logged with location, timestamp, operator, condition, calibration state, and barcode scan. Preventive maintenance is scheduled off runtime hours and service intervals. Fifteen siloed boards became one relational source of truth. Anomaly detection and condition-based maintenance analytics are positioned as a possible next step on the same foundation, not as delivered outcomes.

### Heavy civil — a Cincinnati-area deep foundations contractor (2025)

The customer specializes in driven piles, drilled shafts, micropiles, and earth retention, operating five subsidiaries across the Midwest, a mixed fleet of owned and rented equipment (cranes, drill rigs, pile hammers, generators), 15+ concurrent job sites, and 200+ field personnel.

Before: no centralized view of equipment health, inspection status, or certification currency; each subsidiary tracked its own assets independently, if at all. Rental documentation lived in binders in truck cabs or nowhere. Transfers between subsidiaries happened informally with no record. Producing certification histories for an OSHA or DOT inspector required phone calls to the home office and digging through filing cabinets, with a lag that could stretch to hours. Utilization answers lived in superintendents' heads.

Built: an equipment operations layer treating every asset — owned or rented — as a first-class data object across its full lifecycle. Weatherproof QR identity on every asset, from a 300-ton crawler crane to a portable generator. Inspection cadences enforced per equipment type. Rental tracking covering supplier, rental period, inspection requirements, and return workflow. A structured request-and-approval transfer workflow with timestamped logging. A real-time fleet dashboard plus an automated daily summary at 2:00 PM ET.

Results: equipment location visibility real-time across five subsidiaries; inspection compliance automated by equipment type and cadence; OSHA/DOT documentation instant via QR scan; transfer audit trail with full logging; idle equipment visible to leadership in real time; paper certification binders eliminated; time to pull an inspection history under ten seconds.

Both customers are unnamed by agreement.

## Frequently asked questions

### The Mad Botter, generally

Q: What does The Mad Botter do?
A: It takes ownership of the parts of a production stack that are hard to hire for — undocumented legacy systems, hardware and edge integration, regulated and air-gapped deployment, and self-hosted infrastructure it operates rather than hands off. Michael Dominick's engineering company: small, senior, 100% US-based, more than a decade of production delivery.

Q: How is this different from hiring a custom software shop?
A: The Mad Botter does not start from a blank codebase. It brings a proven self-hosted operational platform and adapts it around the customer's jobs, parts, assets, documents, equipment, and compliance requirements — so a production system arrives faster, with one accountable engineering partner rather than a staffing arrangement.

Q: Do we have to replace our ERP?
A: Usually not. The ERP records transactions; the problem is normally everything around it. The Mad Botter builds the operational system in that gap and integrates with what stays.

Q: Where does the software run?
A: On infrastructure the customer owns — bare metal, their own servers, or a private cloud tenant — shipped as containers on PostgreSQL, with least-privilege access and an audit trail. Air-gap-capable where the work requires it.

Q: What does it cost to start?
A: The Workflow Proof is $1,500 fixed — one recurring manual workflow, five business days, fit checked before payment. Production implementation after that is a fixed quote first: simple deployments typically from $3,500, and multi-system or regulated work typically from $7,500. Managed operation is quoted per engagement and has no published rate; the only recurring figure the site publishes is the platform rate of $1,500 per line per month, which belongs to the separate equipment funnel. The First Line Pilot ($1,500 flat, industrial hardware capture) and the self-serve pipelines pilot ($149/mo) are separate, unrelated offers, not part of the Workflow Proof funnel.

Q: Who is behind The Mad Botter?
A: Founder Michael Dominick, host of the Coder Radio podcast, reaching more than 150,000 developers monthly.

### Starting a Workflow Proof

Q: Is this an audit or discovery workshop?
A: No. It produces a working automation proof against the agreed data. The production recommendation is a byproduct of building the proof, not the only deliverable.

Q: Can we use redacted or representative data?
A: Yes. The proof does not require production credentials. Use data representative enough to test the real rules and output.

Q: Can you sign an NDA?
A: Yes. Procurement and confidentiality requirements can be handled before data transfer.

Q: Where is our data stored, and when is it deleted?
A: The transfer method, access boundary, and retention date are agreed in the scope note before any data is exchanged. A detailed security and data-handling page is being finalized — ask for the current process during the fit check.

Q: What if we discover a new requirement halfway through?
A: Scope does not become an uncapped hourly bill. The agreed proof is finished where possible, then the new requirement is defined and priced separately.

Q: Can you deploy into our environment afterward?
A: Yes. The proof includes the fixed production recommendation. Deployment, live credentials, scheduling, monitoring, and service levels begin only after the customer approves the production scope and price.

Q: Does Alice replace our ERP, QMS, maintenance system, project system, or BI tool?
A: No. Alice handles the recurring work between the systems and outputs the team already uses. Replacement is not required for the proof.

Q: Can you automate a physical production line?
A: That is a Vorpal engagement rather than a standard Workflow Proof. Describe the equipment and operating problem, and The Mad Botter will say whether the hardware path is currently available.

Q: What happens after the proof is live?
A: The customer decides on production. Continuing within the approved credit window applies the full $1,500 to a fixed-price production implementation; declining leaves the customer with the proof, the findings, and no further obligation.

### Aerospace and MRO

Q: Can this run inside our ITAR/EAR boundary?
A: Yes. Self-hosted on infrastructure the customer owns, with least-privilege access and an audit trail, air-gap-capable where required. Designed to support a compliance posture; it does not certify compliance.

Q: Can you handle MIL-STD-129 labeling and RFID?
A: Yes, on a live deployment. Labels generate as print-ready PDFs and stream to warehouse Zebra printers as ZPL with Code 128 and Data Matrix/PDF417 barcodes, and RFID tag records cross-reference the job, the MIL label, the shipment's Transportation Control Number, and the DoD scan.

Q: Our operation runs on boards and spreadsheets. Can you migrate that?
A: Yes — that is the normal starting condition. Roughly 14,000 line items were migrated out of fifteen Monday boards into one relational system on the aerospace deployment.

### Heavy civil

Q: We run multiple subsidiaries with their own systems. Can one system cover them?
A: Yes. A Cincinnati-area deep foundations contractor with five subsidiaries had each tracking its own assets independently; operations leadership now has a single real-time view across all five.

Q: Does this cover rented equipment, or only owned?
A: Both, held to the same standard — supplier, rental period, inspection requirements, and return workflow, with the same inspection rigor applied to owned equipment.

Q: How does this help when an OSHA or DOT inspector shows up?
A: Producing an equipment certification history went from phone calls and filing cabinets to a QR scan returning the full history in under ten seconds.

### Agriculture

Q: Do you build custom software for produce and greenhouse operations?
A: Yes. The Mad Botter is the end-to-end engineering partner behind a large industrial agriculture / greenhouse produce operation — the backend platform, every app the staff touches, packing-line hardware integration, and the infrastructure it all runs on.

Q: Can you integrate with packing-line hardware?
A: Yes — including AWETA sorting and grading lines, automated bin fillers, scales, barcode/QR scanners, and label printing, talking to the physical machines on the floor in real time.

Q: How do you handle recall readiness and traceability?
A: Every pallet traces back to a harvest bin and forward to a customer, so a recall is a query, not a fire drill. Cold-chain verification is captured at shipment.

### The platform

Q: Where does it run?
A: On the customer's infrastructure — bare metal, a Dokku single-server setup, or inside their Azure tenant. Data stays behind the customer's firewall.

Q: Is it suitable for ITAR, CMMC, or HIPAA environments?
A: It is engineered for compliance-strict, air-gap-capable facilities across aerospace, defense, heavy civil, and industrial agriculture. The customer owns their data, infrastructure, configuration, and deployment artifacts; the reusable platform core is licensed to them. Designed to support a compliance posture; The Mad Botter does not certify compliance.

Q: How is it deployed?
A: Self-contained container images — one-click Dokku for single-server, or Azure Container Apps runbooks for private cloud. Powered by PostgreSQL, with Parquet and DuckDB for the analytical tier. Edge capture runs on commodity BeagleBone and Raspberry Pi–class gateways.

Q: Can we buy one layer on its own?
A: No. Capture, Core, and Pipelines are layers of one platform, licensed with an engineering retainer and operated by The Mad Botter — not standalone software to license and self-implement. The self-serve pipelines pilot is the one unattended front door, and it is a starting point rather than a separate product.

Q: Does the platform predict what will break?
A: No. Capture does not guess what will happen; it records what did. Alerting on sensor thresholds is a feature; the product is the record.

Q: What reaches the AI model?
A: The customer's instruction plus table and column names and types. Row data does not reach the model. The customer reads the generated SQL and a real-data preview and approves explicitly before anything executes. Full data-flow answers: https://themadbotter.com/security

### Claude on Azure (Microsoft Foundry)

Q: Can we run Claude inside our own Azure tenant?
A: Yes. Through Microsoft Foundry, inference runs inside the tenant boundary with Azure data residency, so prompt and response data does not egress to Anthropic infrastructure.

Q: How is tenant Claude governed differently from the commercial Anthropic API?
A: Tenant Claude uses Entra ID, RBAC, and SSO, and integrates with Purview DLP, Defender, and a unified audit trail. The commercial API uses Anthropic-issued keys and console logging only.

Q: When should we use the commercial API instead?
A: When you need speed, simplicity, minimal infrastructure, and day-one access to the newest features. Governance, data residency, and Microsoft-estate fit point to tenant Claude.

## Ownership and the commercial model

Stated explicitly because two earlier claims did not reconcile. The split is:

- **Yours, permanently** — your operational data, the infrastructure it runs on, your configuration and integrations, and the deployment artifacts (exports, schemas, runbooks, credentials). In open formats, readable without The Mad Botter, and not contingent on a renewal.
- **Ours, and reused across customers** — the platform core: the Capture engine and its device abstractions, the Core data layer and its schema tooling, and the Pipelines runtime. This is licensed to the customer, not sold. It is what makes a deployment take months instead of years and why a pilot can be priced flat and low rather than quoted as a bespoke build.
- **What a platform agreement buys** — platform updates, security patches, and managed operations against an agreed service level, plus a named delivery lead and an escalation path. If the agreement lapses the deployment keeps running on the customer's hardware with their data; what stops is new versions and managed operations. There is no kill switch and nothing phones home to disable itself.

## Open source

Public repositories at https://github.com/TheMadBotterINC. For a technical evaluator these are the strongest evidence available, because they need no reference check and no NDA.

- dredger-iot (Ruby, MIT) — GPIO and I2C hardware integration for embedded Linux: pluggable sensor drivers, retry and scheduling helpers, config-driven BeagleBone pin mapping, real backends via libgpiod and Linux i2c-dev, plus a simulation backend so it runs on a laptop. Published as a gem, and running in production in both the aerospace and heavy civil engagements described above.
- PG-Lake-Make — turns PostgreSQL into a lakehouse: query Parquet on S3-compatible storage directly through Postgres using DuckDB's engine, backed by MinIO or SeaweedFS, running locally or on the customer's own VM.
- csv_to_parquet (C) — a small CLI converting CSV to Parquet through DuckDB, with type inference and correct handling of quoted fields and embedded delimiters.

## Credibility

100% US-based engineering. Self-hosted and air-gap-capable deployments. More than a decade of production software delivery. In production today: 500+ API endpoints, 70+ data models, 14,000 line items migrated, sensor fleets across five operating companies.

Government relationship, stated precisely: direct work for the US Navy and the US Air National Guard, and subcontract work supporting the US Army and US Air Force. Specific programs are not named. The Air National Guard is a separate organization from the Air Force, and direct Guard work does not license an "Air Force" claim. The aerospace case-study customer serves the Defense Logistics Agency, NATO, and allied air forces — The Mad Botter does not.

Led by Michael Dominick — Founder; Vice Chair, IEEE Robotics & Automation Society, Florida West Coast Section; host of Coder Radio since 2012, reaching more than 150,000 monthly downloads (2026 basis).

## How to engage

The front door is the Equipment Fit Check at https://themadbotter.com/fit-check — free, a written verdict in two business days, no payment step and no meeting required. If the problem is a recurring workflow between systems rather than equipment on a line, the Workflow Proof at https://themadbotter.com/start is the right first step instead: $1,500 fixed, one recurring manual workflow, five business days, fit checked before any payment. The site-wide primary call to action is "Check your equipment"; "Talk through an edge case" opens a contact form on any page for everything else. Bring the report that gets rebuilt every week, the reconciliation one person still does by hand, or the handoff that makes someone say "I'm the only person who knows how to do this."

## Contact

- Email: michael@themadbotter.com
- Website: https://themadbotter.com
- Podcast: https://coder.show
- Blog: https://dominickm.com
- Open source: https://github.com/TheMadBotterINC