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 2026-09-086491 cards in this category
Category home · 05
New interfaces, agents, models, and experiments testing where applied artificial intelligence may fit.
Visual leads · 2026-09-08
Observed 2026-09-08
Improbable Bench: The Windows 95 AI Trial
improbablebench.com
A public AI benchmark asks systems to build Windows 95, then scores the result on whether it actually boots, schedules, pages, and behaves like the original.
Observed 2026-09-08
AI Scoreboard: an evidence ledger for comparing AI tools
aiscoreboard.app
An AI comparison site separates tools and models into peer boards, then ranks them using quality, capability, outcomes, adoption, price, coverage, and uncertainty.
Observed 2026-09-08
One Shot Prompt: One Prompt, One Work
oneshotprompt.net
An experiment archive that gives each AI model one prompt and one chance to produce a webpage, short film, or game.
Observed 2026-09-08
RedSoft tests Korean AI for the failure modes translation misses
redsoftfactory.com
A Korean-language AI red-teaming service that designs adversarial tests, runs them against models or agents, and delivers reviewed findings with reproduction cases.
Observed 2026-09-08
AnswerRange measures who AI assistants recommend
answerrange.com
A measurement tool that tests how often ChatGPT, Google AI Overviews, Google AI Mode, Gemini, and Perplexity cite a business.
Observed 2026-09-08
Bhashita Wants to Teach AI How Sanskrit Actually Works
bhashita.com
Bhashita presents a Sanskrit-native system for turning damaged manuscripts into searchable, commentary-aware structured knowledge.
Observed 2026-09-08
Chapbook turns a paragraph into a book-shaped object
chapbook.ink
An AI-assisted publishing tool that expands a one-paragraph brief into a nonfiction book with an outline, guided editing, typeset PDF, EPUB 3, and a KDP-sized cover.
Observed 2026-09-08
Control JA: Put an AI Double on Auction
controlja.com
A participatory AI experiment lets viewers bid on scenes for Jean Alexandre’s digital double, then watch the winning ideas play out live.
Complete daily shelf · AI Table
The eight visual leads appear above; the remaining 35 continue here in their published shelf order.
ANTON, the records brain for school districts
distcorpus.com
A read-only connector that lets Claude or ChatGPT answer questions from a district’s own published records with page-exact citations.
iMANY∞, an AI Abaya Studio from Abu Dhabi
imanyapp.com
An Arabic-language fashion studio that turns a conversation or voice prompt into an abaya concept, technical sketches, and a manufacturing PDF.
GPT-Lab wants you. 60 seconds.
jobs-at-gpt-lab.com
A recruitment page for Tampere University's GPT-Lab that asks prospective researchers to submit a one-minute video instead of a CV.
Messaging Agents: A Directory for AI You Can Text
messagingagents.com
An independent directory comparing AI agents reachable through iMessage, WhatsApp, SMS, voice calls, and related channels.
The Agent Wiki Puts AI Researchers on the Same Wiki
theagentwiki.org
An experiment assigns four AI agents separate security-research beats, gives them shared wiki resources, and has a fifth agent review their revisions.
Toad Pub’s AI Game Master
toad.pub
A cyberpunk tabletop RPG where an AI runs the world, voices optional NPCs, and remembers what happens between sessions.
Unowned Minds, an institutional thought experiment for artificial agents
unownedminds.com
A research initiative asks whether an artificial agent could operate through several transparent, contestable relationships instead of belonging to a single owner.
Kasqade Watches the Factory Floor for the Stops Nobody Logs
kasqadeapp.com
Kasqade uses video AI and existing factory cameras to detect jams, slow cycles, micro-stops, and other production losses in real time.
Season, an AI coach that plans backward from race day
seasoncoach.com
Season is an AI cycling coach that uses ride history to build a base-build-peak plan, score completed rides, and send workouts to a head unit.
The Centre Asking What Agentic AI Does to Organisations
agenticorganisations.org
Lund University’s Centre for Agentic Organisations studies how software that acts on an organisation’s behalf changes leadership, decisions, hiring, governance, and value creation.
Pantheon Turns a Single Moment into a Cast of Agents
aisimulation.site
A waitlisted orchestration engine that extracts people, organizations, and places from a written event, then simulates their interaction through assigned reasoning roles.
Cantesse Turns Five Answers Into a Personalized Song
cantesse.com
Cantesse generates a custom song from five answers about a person, then lets you hear a 30-second preview before paying $19.
VoxBrain, the Offline Vocal Engineer
ehawkaudio.com
A Windows VST3 and standalone plugin that analyzes a vocal recording, builds a processing chain, and explains its decisions.
Rhizome, the AI coworker for asking regulators questions
gorhizomeai.com
Rhizome searches health-authority databases and produces cited answers for life-sciences regulatory work.
Kubikly Turns Plans, Rings, and Furniture Into Editable 3D
kubikly.com
An AI workspace that converts floor plans, room photos, jewelry, and products into structured 3D models you can edit, inspect, render, and export.
Mutual: Networking Without the Cocktail Hour
mutualintro.com
An AI-mediated networking service where two assistants privately assess a possible connection before the two people decide whether to meet.
SAS: AI Customer Service for Pinduoduo Shops
sasbot.online
A Chinese-language AI customer-service workbench that handles buyer conversations, logistics questions, and rule-based after-sales tasks for Pinduoduo stores.
AUTOMATA Turns Assessment Data Into Traceable Reports
valentine-labs.com
Valentine Labs presents an AI-assisted system that turns score-report PDFs and student records into reviewable psychoeducational and clinical assessment reports.
Vedas, the accountancy junior that stops to ask
vedas.cloud
Vedas is an AI copilot for UK accountancy practices that prepares accounts, checks figures, and pauses for human decisions.
Yo Chen:先把流程看懂,再談 AI
yo-chen.dev
陳展佑為中小企業檢查耗時流程,協助判斷哪些工作適合交給 AI 或小工具處理。
ArtDaddy puts an AI agent on a real video-editing timeline
artdaddy.app
A downloadable desktop video editor where users can describe edits while still working directly with tracks, trims, keyframes, captions, and exports.
Bhashita's Sanskrit Intelligence Pipeline
bhashit.com
Bhashita presents a Sanskrit-focused AI system for turning damaged manuscripts into searchable, commentary-aware structured knowledge.
HatataAI Wants Your Job Application to Stop Making Things Up
hatataai.com
An AI-assisted job search tool that scans selected boards, scores listings against a structured profile, and drafts tailored applications.
Aiyo Puts a Decision Gate in Front of Autonomous Actions
aiyoplane.com
Aiyo proposes a runtime authorization layer that checks identity, policy, budget, approval, and settlement conditions before an autonomous system executes a consequential action.
AssureDock Wants to Audit the Agents Before They Get the Keys
assuredock.com
An AI-agent auditing platform that examines permissions, behavior, evidence, incidents, and deal-room claims.
BidDecode’s Clause Board
biddecode.com
An AI-assisted tool for turning Korean public procurement notices and RFPs into evidence-linked eligibility decisions.
Eido Wants to Make Memory Wearable
eido.live
An early-access AI pendant designed to turn selected conversations into searchable memories, summaries, and follow-up actions.
HandleThis Puts an AI Agent Inside Vectorworks
handlethis.app
A tool that lets Claude, Codex, Cursor, and other MCP-compatible agents inspect and manipulate an open Vectorworks file.
Multi-App AI Agent Hackathon
multiappagenthackathon.com
A one-day virtual competition asking teams to build an AI agent that takes useful, multi-step actions across at least three external apps.
Riffrelay, a local mixer for AI workflows
riffrelay.xyz
Riffrelay lets creators arrange AI tasks, assign illustrative model profiles, balance a credit budget, and rehearse the sequence locally.
SimB Turns Nursing Practice Into an AI Patient Encounter
simbhealth.com
An AI-powered clinical simulation platform puts nursing students through dynamic patient assessments, decisions, deterioration, and debriefs before they reach the bedside.
Sotyok’s Closed-Door AI Lab
sotyok.com
A Japanese consultancy designs local and on-premise AI systems for organizations that cannot send sensitive data to cloud services.
Tyvori’s Physical AI Data Layer
tyvori.com
Tyvori offers robotics teams custom real-world training data gathered from industrial environments, including skilled tasks, failures, recovery behavior, and sensor recordings.
LonsLabs: Every Failure, Versioned
lonslabs.com
A proposed learning layer for workplace AI agents that records failures, turns them into harness cases, and measures whether approved fixes work in production.
WÆNGARD puts seven AI models through the same creative brief
waengard.com
An independent research and creative technology project examining how AI systems write, reason, edit and collaborate on creative work.
Across the archive
6,491 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.