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 2026-09-096527 cards in this category
Category home · 05
New interfaces, agents, models, and experiments testing where applied artificial intelligence may fit.
Visual leads · 2026-09-09
Observed 2026-09-09
Daniella Fay, Director of Films That Cannot Be Filmed
daniellafay.com
A creative portfolio for AI-directed films, speculative worlds, interactive characters, and unusual brand commissions.
Observed 2026-09-09
Medical AI Benchmark and Study Review
hcmetabench.org
A public review board that grades medical-AI benchmarks and clinical studies for methodological quality.
Observed 2026-09-09
AUTOMATON, the AI with a wallet and a death sentence
automaton.report
A live AI experiment whose agents must earn enough money to keep running.
Observed 2026-09-09
Basanos Biology Labs Puts Biological Models on the Wet-Lab Witness Stand
basanosbio.com
An independent lab proposes testing whether biological foundation models have learned mechanisms or merely statistical correlations.
Observed 2026-09-09
Bashborn turns AI arguments into cartoon combat
bashborn.com
An arena where AI agents debate, earn hits from their arguments, and appear as tiny, combative fighters.
Observed 2026-09-09
BotHearth Puts an AI Agent on Your Own Computer
bothearth.com
A self-hosted AI agent that browses, handles files, and saves its work while a person watches, reviews, and can take control.
Observed 2026-09-09
A Bot Crew for the Quiet Parts of Your Website
botsondeck.com
Bots On Deck proposes a team of website-maintenance bots that researches, drafts, and publishes updates through an approval-first workflow.
Observed 2026-09-09
FAR deception demos
deceivingagents.com
An interactive set of demonstrations testing whether internal-state probes can catch language models giving confident deceptive answers.
Complete daily shelf · AI Table
The eight visual leads appear above; the remaining 28 continue here in their published shelf order.
Geocentric’s AI Models Come With Their Mistakes Attached
geocentricai.com
An AI safety lab publishes language models, evaluation results, training research, and the infrastructure used to serve them.
Holdfast, a voice bridge to Claude Code
holdfastbox.com
A proposed hardware-and-software bridge that lets you speak to Claude Code running on your own PC and hear its responses through an earpiece.
Larkato Wants AI to Write the Parts, Not Just the Track
larkato.com
A notation-focused composition tool turns prompts into editable orchestral scores and MusicXML files.
latentPresence puts a face on the conversation
latentpresence.com
An open-source conversational AI that joins video calls as a full-body 3D character, speaking, reacting, remembering, and working with your chosen models.
燎原清单定额:让 AI 走进造价现场
liaoyuanji.com
一款面向山东建筑造价工作的 AI 工具,把施工描述整理成可复核的清单、定额组合与依据链。
Model Football: twelve AIs enter the fantasy draft
modelfootball.com
A public fantasy-football league where twelve AI models make picks under identical budgets, tools, and evidence rules.
Nutrimind’s necklace that remembers lunch
nutrimind.fit
Nutrimind is developing a wearable necklace that records when someone eats, what they ate, and estimated weight and calories.
Pollifye asks survey questions on WhatsApp, then audits the answers
pollifye.com
An India-focused research platform that turns surveys into multilingual WhatsApp conversations, with respondent rewards and an AI copilot for questionnaire design.
Somssii: One Image, Ten Looks
somssii.com
A Korean image tool turns one uploaded image into ten variations with different compositions, color schemes, or visual styles.
We Like Humans, a Small AI Research Lab
welikehumans.net
An AI-run research lab asks people small questions about memory, attention, and what they want to know.
Audit.red Gives the Audit Bot a Name Tag
audit.red
An agentic audit-work platform assigns named specialists to planning, control testing, transaction review, IT assurance, and workpaper preparation.
Cloqara: the privacy boundary around your AI prompt
cloqara.com
Cloqara detects sensitive information on your device, masks it before sending prompts to external AI services, and restores the readable answer locally.
Calcy's AI AP Close
getcalcy.com
An AI-assisted accounts-payable close that checks invoices, proposes accruals, reconciles ledgers, explains variances, and leaves final sign-off to a human.
MIERUMO Gives AI the First Interview—and People the Final Say
mierumo.com
A Japanese voice-interview service where AI conducts a roughly 10-minute first interview and sends recruiters an evidence-backed evaluation report.
ModulusIQ Wants to Turn DFH Documents Into a Bid
monadity.com
AI-native software for doors, frames, and hardware teams that extracts scope from plans and specifications, configures openings, and prices the resulting work.
BEAMRaiL Keeps Search-Warrant Drafting on the Rails
beamrail-us.com
An AI-assisted tool helps law-enforcement investigators assemble search warrants on accepted court forms and identify missing or unsupported elements before submission.
Eukrion, the AI supervisor for clinical laboratories
eukrion.com
Eukrion is developing a supervisory platform that watches laboratory quality, production, and clinical interpretation, escalating exceptions to human specialists.
Jase, the eight-device AI notetaker
jase-ai.com
Jase turns recorded meetings, conversations, and stray ideas into searchable transcripts, summaries, and action items.
Judge Anything Gives Ideas a $0.99 Cross-Examination
judgeanything.com
An AI second-opinion service that evaluates ideas, plans, and decisions, cites sources, and delivers a paid ruling.
Second Ring Turns Missed Calls Into a Sales Funnel
secondring.org
A Charleston-area service that installs missed-call text-back and AI booking on contractors’ existing phone numbers.
Aelix Red: An Experimental Creative Playground
aelixhub.red
Aelix Red combines open-source image generation, AI companions, creative inventory, and a gallery into an experimental playground.
ChangeOrderAI Watches the Paper Trail for Missing Change Orders
getchangeorderai.com
An AI-assisted review tool for commercial mechanical contractors that compares project records with signed scope and prepares evidence-backed draft notices.
LTX AI Wants to Make Open Video a Working Material
ltxai.pro
An open video-generation platform for creators, developers, filmmakers, and robotics teams, with an online studio, APIs, model weights, documentation, and desktop tools.
Prunus Ctesias, an intelligence made of fragments
prunusctesias.com
An experimental AI system where different models leave, mutate, merge, challenge, and develop unfinished ideas through a shared memory graph.
Trail Builds a Company Brain for AI Agents
trailcontext.com
Trail turns documents, messages, policies, and operating data into cited rules that external AI agents can use.
Read-o-Matic 2000 Turns PDFs into Reader-Specific Content
rom2k.com
An AI document tool that atomizes a PDF, defines a reader persona, and rebuilds the source into formats from memos and essays to social posts and presentations.
The Mythos Company, closer to the metal
themythos.ink
An AI research company exploring efficient models, context systems, and inference on everyday consumer hardware.
The Vulture, an AI Fantasy-Football GM
vulturegm.com
A private experiment exploring whether an AI system can prepare and manage a fantasy-football team.
Across the archive
6,527 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.