← The AI Hype Audit — all 524 verdicts

PARTLY

Partly true: the DIY 'Jarvis' is a real, working personal-assistant setup (offline transcription, a note vault, autonomous agents on a Mac Mini and a Raspberry Pi), but calling it as smart as Iron Man's Jarvis is the creator's own stretch, which he half-admits

Facebook · Oct 2026

The claimI'm building my own Jarvis. A Plaud recorder captures everything I say, an offline LLM transcribes it, a big 'map' sorts it for my agents, skills keep the loop updating. Hermes agents run on a Mac Mini with their own computer; Fletcher on a Raspberry Pi speaks to me via an offline LLM and pulls knowledge from my second brain. People say AI isn't even as smart as Jarvis, so why try; relevance of the memory is what makes it feel intelligent.

The build is real and technically coherent, not a fantasy. Capturing audio with a wearable recorder, transcribing it with an offline (local) model, storing it in a linked note vault as a 'second brain,' and running autonomous agents on a Mac Mini with a Raspberry Pi as a voice front end, pulling knowledge on demand, is a plausible, if elaborate, personal-assistant stack, and every component named (local transcription, a note vault, agent runners, a Pi voice node) exists. Where it tips into hype is the Jarvis framing. Today's assistants are not Iron Man's Jarvis: no real reasoning-about-the-physical-world, no reliable autonomy, and these pipelines are brittle and need constant maintenance. To his credit, the creator half-concedes this, saying it works because the memory is relevant to him, not because it is superintelligent, which is an honest reframing. So treat it as an inspiring, genuinely functional DIY project, not a blueprint for a movie AI. Real stack, aspirational name, honest narrator.

What holds up

  • Reel transcribed in full and caption read.
  • Confirmed every named component is real: wearable recorders, offline/local transcription models, linked note vaults, agent runners, Raspberry Pi voice nodes.
  • Noted the creator himself reframes 'Jarvis' as relevance-of-memory, not superintelligence.

What doesn’t

  • 'Jarvis' framing oversells today's assistants.
  • The pipeline is brittle and maintenance-heavy.
  • It is one person's custom build, not a reproducible product.

The catch

The setup genuinely works; it is a well-organized memory assistant, not a reasoning movie AI.

How to actually do it

  • If inspired, start small: local transcription plus a linked note vault before adding agents.
  • Expect ongoing upkeep; these multi-device pipelines break often.
  • Judge it by whether it recalls what you need, not by the Jarvis label.

A real, working DIY assistant with an aspirational name the creator himself tempers.

Confidence
Medium
Posted by
a maker/AI creator page (about 22,000 plays); reel sent to Buddy Thu Oct 15, 2026

See the original claim →

We test hype for free. We build the real thing for a living.

Thirty minutes, no pitch — and you'll leave with something useful either way.

Book a call with Todd or start with the free Business Checkup →

The Verdict Weekly

Three verdicts every Friday. Free forever, unsubscribe anytime, no spam — that would be ironic.

© Schreier Group · schreiergroup.com · See a wild AI claim? Drop it here and we'll test it.