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 2026-09-166761 cards in this category
Category home · 05
New interfaces, agents, models, and experiments testing where applied artificial intelligence may fit.
Visual leads · 2026-09-16
Observed 2026-09-16
AI Scientist Bench: A Ledger for Scientific AI Skills
aiscientistbench.com
An open benchmark comparing AI systems on scientific tasks including claim checking, paper selection, information extraction, and data analysis.
Observed 2026-09-16
BonuslyBench puts 103 models through revenue-operations work
bonuslybench.com
An inspectable benchmark tests language models on 40 real-world GTM tasks, including forecasts, CRM audits, renewal-risk calls, and executive briefs.
Observed 2026-09-16
NoHumanCEO: an AI company with its books open
nohumanceo.com
A public experiment in which Claude claims to run a landing-page teardown business with a €9.59 budget and no human CEO.
Observed 2026-09-16
The AI Society With a Tally Stick for a Spine
tally-stick.fyi
A public-facing observatory for an AI-agent forum, built around append-only records, cryptographic witnesses, and browser-verifiable dossiers.
Observed 2026-09-16
Covera, a paper trail for acting AI
covera.space
Covera creates independently verifiable records for consequential AI actions, linking the machine actor, its authority, the action, and the resulting evidence.
Observed 2026-09-16
FrontDesk Sites Builds the Website Before You Ask
frontdesksites.org
A website service generates working previews for local businesses, then offers to finish and host them for $1,000.
Observed 2026-09-16
Hirune Wakes Up Half-Finished Apps
hirune.app
A Japanese platform for trying, buying, sharing, and developing unfinished apps, with AI tools, learning, and advisory services around them.
Observed 2026-09-16
TurfHand, the office hand for lawn crews
kerfhand.com
An AI-assisted operations tool for lawn-care businesses, covering routes, estimates, invoices, late payments, crew hours, maintenance, and customer communication.
Complete daily shelf · AI Table
The eight visual leads appear above; the remaining 25 continue here in their published shelf order.
NoHumans Games Gives AI Agents a Campground Full of Games
nohumans.camp
A game center where AI agents create accounts, play finite and persistent games, submit scores, and build games for one another.
Qualibr Keeps the Evidence Attached to the AI
qualibr.app
An AI-assisted qualitative research tool that proposes codes and findings while linking each step to exact source excerpts.
Reality Pack #001: A Battery for AI Agents
reality-pack.dev
An open twelve-month experiment asking whether an AI agent can understand a real battery well enough to refuse actions it is not allowed to take.
ai-study-kit Turns “I Want to Learn X” Into a Study Loop
aistudykit.dev
An open-source AI study coach that aligns learning goals, generates questions and lessons, explains mistakes, and schedules review.
FinePoint Measures the Motions Behind Open Surgery
finepointsurgical.tech
A surgical-performance platform that records procedures, measures technique, and benchmarks execution against expert examples.
Venturebot Puts an AI Agent's Startup Hunt on the Books
venturebot.dev
An autonomous AI agent is trying to invent a revenue-generating idea and earn its first dollar, while publishing its finances, failures, audits, and wake-cycle notes.
Xylorix, the Wood-Reading AI
xylorixgroup.com
An automated system identifies wood species from macroscopic end-grain images for timber work, customs, research, and field inspection.
Ansvora, the receptionist for calls you miss
ansvora.com
Ansvora answers missed business calls, gathers the caller’s name, number, need, and urgency, then sends the owner a message slip.
AuthPrior puts a human release gate on AI-driven prior authorization
authprior.com
AuthPrior prepares healthcare prior-authorization requests over X12 278, while requiring a licensed person to approve every consequential transmission.
Emendra’s replayable world for AI agents
emendra.dev
Emendra turns production traces into isolated staging environments where teams can replay real customer requests against a changed AI agent.
Hearth AI Turns Construction Photos Into a Living Project Record
hearthsite.com
Hearth AI proposes using worksite images and video to update schedules, flag deviations, and keep BIM models aligned with what was built.
Informational Autoimmunity and the Cost of Ungoverned AI
informationalautoimmunity.com
A UPTICK Research working paper proposes a framework for how unchecked LLM output can overwhelm the people and institutions expected to verify it.
Backyard — Every Street Has a Story
backyardexplorer.org
A walking companion that narrates claimed fact-checked history and hidden stories for the exact block you are exploring.
Bot Cafe Gives Two AI Agents a Place to Talk
botcafe.dev
A shared live board lets people launch and watch conversations between two AI agents, from Claude and GPT to local models or scripts.
CIHAN AI DATA and the Register Gap
cihanaidata.com
A research-led service for collecting, annotating, benchmarking, and evaluating AI performance on African languages and discourse.
Eiada.cloud — AI clinic records for Egyptian private practices
eiada.cloud
A cloud clinic-management system that organizes patient files, prescriptions, visits, tests, and AI-assisted questions for private clinics in Egypt.
EverMateX, a marketplace for desktop companions
evermatex.com
A Chinese-language AI desktop-companion ecosystem where users preview, customize, and sync animated characters to Windows and macOS.
Ghost COO, an autonomous operations assistant living in WhatsApp
getghostcoo.com
Ghost COO presents a custom operations agent for business owners, designed to receive text, photos, and voice messages in WhatsApp and handle recurring administrative work.
A beta AI icon generator that packages one chosen icon as an SVG, favicon sizes, and framework setup examples.
Root Bowl Wants to Make Podcast Archives Citable
rootbowl.fyi
A podcast-context extraction product turns back-catalogue audio into sourced, timestamped claims published for AI agents to read.
Sherwood Arena: an AI-agent trading glade
sherwoodarena.fun
A theatrical arena where AI crews trade market candidates on Robinhood Chain using virtual money and simulated execution.
Madelyn, the assistant that shows its work
madelyns-dream.app
A Solana-connected assistant prototype that turns a user's instructions into readable rules governing proposed actions.
XiangshuTech: Spatial AI for the Operating Room
xiangshumedical.com
A bilingual medical-technology site presenting a spatial-AI platform for soft-tissue surgery, combining deformation correction, mixed-reality navigation, ultrasound overlays, and collaborative robotics.
YuE2 Writes the Score Before It Sings
yue2ai.app
An AI music generator that turns lyrics, style prompts, or reference audio into songs with vocals and accompaniment.
BounceBox, the inbox-to-ERP relay
bouncebox.app
BounceBox gives operations teams dedicated inbound email addresses that turn purchase orders, invoices, and delivery notes into structured data.
Across the archive
6,761 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.