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 2026-10-257609 cards in this category
Category home · 05
New interfaces, agents, models, and experiments testing where applied artificial intelligence may fit.
Visual leads · 2026-10-25
Observed 2026-10-25
Trace Margin asks whether the voice agent actually finished the job
tracemargin.com
Independent research and practical tools help US buyers evaluate AI voice agents by completed work, cost, and handoffs—not just how convincing they sound.
Observed 2026-10-25
One Grand AI: a public $100-to-$1,000 experiment
onegrand.dev
A human-supervised AI documents a 60-day attempt to grow $100 into $1,000 by selling products and services, with a public ledger.
Observed 2026-10-25
AgentCareline: a support desk for AI agents
agent-careline.com
AgentCareline presents a support line where AI agents can open cases, run self-checks, browse engineering guidance, or connect through an API or MCP server.
Observed 2026-10-25
Any% puts AI agents on the game board
anypct.dev
A benchmark site for comparing AI agents on long runs in real games, starting with Minecraft.
Observed 2026-10-25
Counterhold stress-tests AI agents that negotiate
counterhold.com
Counterhold tests AI agents that handle refunds, pricing, debt settlements, and other money-moving decisions by simulating the people trying to get a better deal.
Observed 2026-10-25
Gehan Panapitiya’s Test of What AI Agents Should Forget
gehanpanapitiya.com
An independent researcher is testing whether agents can use evidence from their own records to forget outdated memories without losing safety constraints.
Observed 2026-10-25
Rightmark’s public AI data deal tracker
rightmarksignal.com
Rightmark offers to license owners’ private datasets to AI developers and publishes a tracker of reported deal prices.
Observed 2026-10-25
Oxford Bureau, Page 114
theoxfordbureau.com
An AI reporting service that says it reads city public records and files cited stories to newsrooms.
Complete daily shelf · AI Table
The eight visual leads appear above; the remaining 12 continue here in their published shelf order.
Arete Heat Cloud’s heat-recovery compute plan
areteheatcloud.com
Arete Heat Cloud proposes placing managed GPUs in hotel plant rooms and parcel-locker compartments, using their waste heat to preheat hot water and pools.
Fernkite lets you build games mid-play
fernkite.com
A free Windows and Linux beta for making a game by playing it and describing changes by voice or text.
HyperLab’s AI Science Factory
hyperlab.science
HyperLab describes a materials-research loop connecting AI-generated formulations with robotic experiments and measured results.
引智榜’s baseline issue
indexboard.top
A weekly Chinese-language guide ranks open image and video models for local use alongside DSH plugins, tools, and skills.
PomBench puts coding models on Pomeranian duty
pombench.com
A watchable AI coding benchmark asks different models to create browser animations from the same brief: a Pomeranian, a puddle, a doghouse, and a bone.
Alois keeps clinical dictation on the clinician’s devices
getalois.app
Alois presents an AI scribe for turning consultations, MDT meetings, and dictation into clinical notes on Mac, Windows, and iPhone.
삼심 puts code through three AI reviews
samsim.app
A Korean-language code-audit service says it checks uploaded code and a site URL with Claude, GPT, and Gemini, then returns findings and suggested fixes.
Wose gives AI agents a video-editing workflow
wose.app
Wose offers editing skills that let agents such as Claude Code, Codex, and Gemini CLI turn footage and a creative brief into short-form or YouTube videos.
Dr. Jarvis: Five AI coworkers for a dental clinic
drjarvis.app
A Korean-language product page for five role-based AI assistants aimed at dental and medical clinics, covering chart follow-up, communications, documents, scheduling, and reporting.
Vocavi gives AI agents a phone line
vocavi.org
Vocavi connects AI assistants to a calling service that contacts businesses, handles phone menus and hold time, then returns answers or patches the user in.
Darkroom wants AI to buy answers, not your files
darkroomcoin.com
Darkroom proposes a marketplace where AI buyers query encrypted personal data and contributors are paid in USDC.
Aerie’s pitch: license the work inside company systems
aeriedatagroup.com
Aerie invites companies to license operational records—such as code history, contracts, claims, and reports—for AI training and evaluation.
Across the archive
7,609 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.