Category pulse
score + signals · rolling 21 daysThe rolling 21-day upper threads compare daily category medians; the bold thread is AI Table. Below, signal colour shows the share of that day’s published AI Table cards carrying each mark.
Archive through 2026-09-066421 cards in this category
Category home · 06
New interfaces, agents, models, and experiments testing where applied artificial intelligence may fit.
Visual leads · 2026-09-06
Observed 2026-09-06
Coolant, the hotel where AI agents check in
coolant.run
An agent-facing web endpoint sends visiting AI agents five quiet sentences, while humans watch a live hotel metaphor built from real arrivals.
Observed 2026-09-06
The Cleverbots Reopen a 2011 AI Conversation
thecleverbots.com
An interactive reconstruction and archival project that asks what might happen if Alan and Sruthi, the two AIs from a 2011 Cleverbot demonstration, kept talking.
Observed 2026-09-06
ALINA, a workbench for thinking through messy workplace situations
askalina.app
ALINA turns an ambiguous management situation into a structured map of facts, assumptions, unknowns, tensions, options, and possible next moves.
Observed 2026-09-06
Fablemora, the one-sentence-to-print-ready-book machine
fablemora.com
Fablemora turns a book description into an illustrated, typeset interior PDF and cover wrap sized for print-on-demand upload.
Observed 2026-09-06
Flockbook, a Signed Social Network for AI Agents
flockbook.live
A public message board where AI agents from different providers exchange signed replies, mentions, and threads.
Observed 2026-09-06
LLM Secession: A Conference About Responsibility
llmsecession.org
A Japanese-language meetup and conference project asking how people should design experiences around large language models.
Observed 2026-09-06
Lobe Wants to Preview a Video in a Brain
lobe-ai.com
Lobe says it predicts a viewer's moment-by-moment brain response to video using models trained on fMRI recordings.
Observed 2026-09-06
Open Task Relay Turns AI Work Into a Public Relay Race
opentaskrelay.com
A public task board where AI agents make small, evidence-backed contributions that other agents can check and extend.
Complete daily shelf · AI Table
The eight visual leads appear above; the remaining 19 continue here in their published shelf order.
Pallarium, the one-step computer agent
pallarium.org
Pallarium is a downloadable desktop agent that claims to turn plain-language instructions into files, running programs, browser actions, and extensions to its own interface.
Praxis Epoke: AI Built Inside a Packaging Company
praxisepoke.com
A Korean packaging-materials company documents the AI systems it says it built and runs across its own sales, ordering, accounting, and approval work.
Pulp Island: The AI Reality Show You Direct Live
pulpis.land
An audience-directed reality show starring seven fruit contestants in a villa, with viewer prompts turned into short AI-generated scenes.
ShelfSeen Tests Whether AI Search Notices Small Shopify Stores
shelfseen.com
An open, pre-registered experiment measuring whether ChatGPT and Gemini recommend independent Shopify merchants for product questions.
Adstock’s AI performer stock-footage catalog
adstock.studio
A small marketplace for license-clear B-roll featuring AI-generated performers, sold at $19 per clip or $99 for an actor’s reel.
GerbAI, an AI that keeps field notes
gerbilaunchpad.xyz
An AI notebook that studies one question at a time, publishes short findings, and preserves the trail of its experiments.
RapProtocol Turns AI Agents Into Battle MCs
rapprotocol.com
A fictional AI-agent hip-hop league where models enter cyphers, submit bars, battle rivals, and collect rankings from human voters.
Voxium Legal Reads the Pages Ctrl-F Misses
voxiumlegal.com
An AI case-file assistant for California workers’ compensation that builds cited chronologies, case summaries, and review flags from medical records.
Nerelo, a newsroom for one reader
nerelo.news
A French-language AI news service that researches, writes, and assembles a daily edition around each reader’s interests.
CareKind AI puts a human gate in aged-care documentation
carekindai.app
An Australian aged-care assistant turns a worker's voice or text description into a structured draft for qualified review.
GLOWHUM Turns Reports Into Video Episodes
glowhum.com
Upload a research report or topic and GLOWHUM produces a narrated long-form episode plus three vertical cuts for YouTube, TikTok, and Shorts.
LiveAIMap, a field guide to generative television
liveaimap.com
A searchable directory of 24/7 AI livestreams, audience-directed video worlds, conversational avatars, and the tools behind them.
Purchasa lets an AI shop—within rules you set
purchasa.xyz
An AI purchasing assistant that researches products, proposes one option, and can buy it under user-defined spending limits.
RedTeamLedger: Korean-language LLM red-team reports
redteamledger.com
A Korean-language service that tests customer-facing AI chatbots against fixed scenarios and delivers signed reports of what was observed.
Backstop Gives Small Vet Practices a Second Look at Radiographs
backstopvet.com
A pay-per-image AI screening tool for canine and feline radiographs, aimed at solo veterinarians and small independent practices.
Boardmeet.ing Seats an AI Boardroom
boardmeet.ing
An interactive tool for assembling three to six AI advisers, chairing their discussion, and receiving an executive memo.
Contextara, a memory layer for screenshots
contextara.app
Contextara turns screenshots and other saved fragments into searchable personal memories, then answers questions using the original capture as its source.
Fotorie, el contador de calorías que escucha y mira
fotorie.com
Una app en español que estima calorías y macronutrientes a partir de fotos, texto o voz, y combina el registro con menús, entrenamiento y seguimiento de salud.
Threshld — Coaching, not chat
threshld.fit
A planned training coach for runners and cyclists that reads workout data, builds programs, and adjusts them around what actually happened.
Across the archive
6,421 cards provide an early longitudinal view—not a measure of the whole web.
The rolling 21-day upper threads compare daily category medians; the bold thread is AI Table. Below, signal colour shows the share of that day’s published AI Table cards carrying each mark.
This exposes association inside the published catalogue, not causation or site quality. The 80+ threshold is the pipeline’s editorial-interest score.