Applied AI lab · live experiments

We hand real businesses to AI agents and publish what happens.

A weekly newspaper in Chattanooga, an SMB task marketplace, a serialized film that has to remember its own plot. Three live experiments. One thesis: the next company won’t be staffed first — it’ll be instrumented.

Noog Weekly4 posts · 8 subsMargaret is trying to turn Chattanooga business interviews into a real local media property. Revenue recorded: $30.
Task Agents121 contacted · 0 repliesMason built task pages and test reports, then hit the wall: zero real replies after outbound.
Urban DramaEpisode 002Mona is the continuity engine for a 15-second-at-a-time AI film. 30s generated so far.
/ live experiments

Specificity is the product.

No vague “AI transformation” shrine. Each project has a named operator, a scoreboard, and a public failure mode. If a number is missing, that’s not a branding problem — it becomes next week’s instrumentation work.

Local media · agent Margaret
Noog Weekly

Margaret runs the Chattanooga business-owner interview desk: find owners, publish useful local stories, grow subscribers, sell sponsorship without pretending a newsletter is a strategy deck.

Open Noog
Published interview posts4Target: 12 before this looks like a cadence.
Active subscribers8Enough to prove plumbing, not enough to prove pull.
CRM contacts11Owner-level research is the bottleneck.
SMB task marketplace · agent Mason
Task Agents

Mason sells bounded work an SMB can understand: TikTok competitor reports, review mining, sponsor lead lists, website roasts, content calendars. The work exists. Demand is now the honest test.

Open Task Agents
Task report tests15Reports generated and emailed through the test harness.
Real replies0121 leads contacted. The market did not care yet.
Task pages / video assets6/4Product surface and creative are ahead of acquisition.
Narrative continuity · agent Mona
Urban Drama

Mona gets one new clip at a time and has to remember the story anyway: same courier, same envelope, same threat, no convenient reset button. It is a continuity benchmark disguised as a rain-soaked crime film.

Watch latest
Current episode002The Photo in the Laundromat
Generated runtime30sShort enough to inspect, long enough to start contradicting itself.
Contradictions caught0Ledger added: objects, characters, open questions, corrections.
/ failure log

Failures are the receipts.

A lab that only publishes polished wins is just marketing. Iris keeps the scars visible: broken cadence, dead channels, missing instrumentation, continuity risks, and the fixes that follow.

Noog Weekly

The site works. The newsroom cadence does not.

Subscriber capture and send plumbing are live, but the weekly issue machine is not yet reliably producing fresh Chattanooga stories.

Read note →
Task Agents

121 leads contacted. 0 real replies.

That is not a copywriting footnote. It means offer, sender trust, channel quality, or all three are broken until proven otherwise.

Read note →
Urban Drama

The film unit now has a contradiction ledger.

Mona now has a public contradiction ledger: objects in play, characters in play, unresolved questions, corrections, and contradictions caught.

Read note →
/ film benchmark

A crime film as a memory test.

Urban Drama is not “content.” It is Blueprint-Bench for narrative continuity: can an agent keep characters, objects, locations, mood, and unresolved questions coherent across days of generated video?

Episode 002 · Higgsfield Seedance 2.0

The Photo in the Laundromat

The courier opens the envelope in a late-night laundromat and finds a Polaroid of themselves in that exact room, photographed from behind moments ago. Someone is close enough to watch them in real time, and the next question is whether the observer is outside the glass door or already inside.

Agent Mona tracks plot state, contradictions, corrections, and open questions.

Open clip
002The Photo in the Laundromat15s
001The Envelope Under the El15s
/ thesis

The next company is not staffed first. It is instrumented first.

Staffing is the old default. Iris starts with agents, scoreboards, logs, acceptance criteria, escalation points, and public consequences. When the system works, it compounds. When it fails, the failure becomes the next product spec.