Seidr Labs is the studio name for my work: custom industrial machinery and PLC automation built with my partners at RND Automation, embedded electronics for farms and security, business software running in production, the Minecraft systems people actually use — and a small research practice re-checking real scientific questions. No theory projects, no vapour — just things that work, and questions worked honestly.
Custom-built bagging, screening & packaging machinery for agriculture and mining. I do the SolidWorks design, the Delta PLC ladder logic, the electrical build, and the on-site service.
RND Automation's internet-connected gate-motor controller and anti-theft alarm. Open by call, SMS or dashboard; tamper alerts; OTA updates. ESP32 + GSM on a custom PCB.
Arduino-driven centre-pivot irrigation controllers for farms. Long-range LoRa wireless links across rural sites. Custom carrier PCBs, built to survive the field.
A WhatsApp-native CRM for tradesmen — quote, schedule, invoice, follow up. Built around how trades actually work in South Africa, not how software thinks they should.
Face-scan time & attendance for South African small businesses. Clock in by face (or PIN), live timesheets, payroll export, WhatsApp onboarding. POPIA-built.
A POPIA-compliant, audit-logged incident-record service for licensed South African security firms — a lawful replacement for the informal WhatsApp groups the industry relies on.
Self-hosted vehicle tracking & immobiliser for South African owners and small fleets. Live map, geofences, remote lock — with a custom ESP32 + cellular + GPS tracker.
A Paper plugin that turns any Minecraft server into a managed environment — dashboard, sync, anti-cheat, role system, voice commands for admins.
A Minecraft companion mod — voice-controlled AI that plays beside you. Crafting, mining, building, defense. Designed to feel like a real friend, not a bot.
Alongside the products, the lab runs a small research practice: take a real scientific question, build it into a knowledge map, find the cheap experiment that could actually settle it, run it, then have independent AI reviewers try to tear the result apart before anything is called a finding. The rule is the same as everywhere else here — separate fact from interpretation, cite the real sources, say "contested" when it is contested, and report a null result as proudly as a positive one.
A from-scratch investigation of the "quantum consciousness" idea — built as a living map of the published literature, then a set of cheap, discriminating tests for whether quantum effects could plausibly survive long enough to matter inside neurons. Every experiment was run, then checked twice by independent skeptical reviewers.
Honest verdict: on current evidence the popular "Orch-OR" version is not supported — but not disproven either; genuinely unsettled. The one rigorous, still-testable variant is flagged as the one to watch. Real quantum biology (bird navigation, enzyme tunnelling) is the proven contrast — but it is narrow, and it isn't the brain.
A consistency census of published papers: re-compute every reported p-value from its own test statistic and degrees of freedom, and count how often they don't agree — and how often that disagreement would flip a result from "significant" to "not." This has been done exhaustively in psychology, but barely anywhere else. The lab is taking it to a field that has never had the check.
The early pilot signal is sobering: in the papers scanned so far that report classic test statistics, more than half carried at least one internally inconsistent number, and about one in eight had an error big enough to change the headline result. Full census in progress.
Many celebrated empirical "laws" — the neat scaling relationships that show up in textbooks — were established on a handful of datasets and rarely re-checked against everything that's since come available. Following the playbook that showed "scale-free networks" are far rarer than claimed, the lab re-audits a chosen scaling law across the full set of open datasets it's supposed to govern, with proper model-comparison statistics rather than a fitted line on a log-log plot.
A clean confirmation and an honest "it doesn't hold as widely as we thought" are equally worth publishing. The point is the careful re-check.
In Old Norse, seiðr is the practice of seeing further than the world wants you to — of working with patterns, with foresight, with the long view. That's the spirit of this lab: small, deliberate, durable. Hand-built. Things made to last.
Everything here is made by me — sometimes with the steady help of an AI collaborator (Claude) at the keyboard for the software side, sometimes with a multimeter and a torque wrench on the factory floor. Production systems, real money, real users. No marketing spin, no growth-hacks. Just engineering people actually use.