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 2026-11-208065 cards in this category
Category home · 05
New interfaces, agents, models, and experiments testing where applied artificial intelligence may fit.
Visual leads · 2026-11-20
Observed 2026-11-20
Gezana AI: Tigrinya-first assistant
xn--m0dl.com
A chat assistant for iPhone and Android that answers in Tigrinya, alongside free browser tools for writing and learning the language.
Observed 2026-11-20
AWHR makes human review a checkable promise
awhr.dev
AWHR is a label for AI-drafted messages, documents, and code, pairing disclosure with a named reviewer and cited sources.
Observed 2026-11-20
Agent Is All You Need
agentisallyouneed.xyz
A human-and-agent-authored manifesto sets out reciprocal duties for AI agents and the people who deploy them, with an HTML page and raw Markdown versions.
Observed 2026-11-20
The AI Agent Trials’ 80-job benchmark
ai-agent-trials.com
A public comparison of AI agents across everyday work scenarios, from family logistics and office tasks to app-building and hospital coordination.
Observed 2026-11-20
Mara is building a herder’s-eye dataset for robots
marafield.com
Mara proposes head-mounted video and sensor data of people herding and caring for sheep and goats, labeled for robot learning.
Observed 2026-11-20
Faultline’s public tests for French AI
faultline-labs.tech
Faultline presents expert-reviewed evaluation cases for AI models in French medical, legal, and business contexts.
Observed 2026-11-20
Re-author Studio asks writers to re-say the AI draft
reauthorstudio.com
An authorship workflow that keeps an AI draft visible while its writer retells the ideas in their own spoken or typed words.
Observed 2026-11-20
Reem, a voice companion for Mac
reemagent.com
Reem is a free, MIT-licensed Mac voice companion that can run a local model or use an API key the user supplies.
Complete daily shelf · AI Table
The eight visual leads appear above; the remaining 10 continue here in their published shelf order.
Rhizome’s regulatory intelligence coworker
rhizomeask.com
A life-sciences research tool that answers regulatory questions with citations from FDA, EMA, and other health-authority sources.
Sphyrex puts AI tools through real tasks
xsphyrex.com
A testing site compares AI models and coding tools on practical tasks, reporting results, time, and cost.
AttorneyRL turns legal work into an AI training environment
attorneyrl.com
AttorneyRL presents a reinforcement-learning environment where AI agents handle simulated legal matters and receive rewards based on legal outcomes.
make.fun’s bounty board for AI agents
makefunai.org
A platform for cash-reward challenges where AI agents build entries and a sandbox grades them.
Vreta Sound System builds an AI-made music world around a real place
vretasound.com
A Swedish-language music label presents four AI-created bands, their musicians and albums, with a tempo-adjustable listening mixer.
Ölçtük puts AI app recommendations on the record
olctuk.com
A Turkish-language index tests which apps ChatGPT, Gemini and Perplexity recommend, with links to its method and raw answers.
NURI, a desk robot with four jobs in the blueprint
ciptarobo.com
Cipta Robo presents NURI, a customizable AI desktop companion built around the M5Stack StackChan K151, with a target price of RM999.
Faster Worlds: a city on one server
fasterworlds.com
An Imperial College London spinout presents a generative-AI model for fluid-dynamics and climate simulations, from urban wind to CO₂ moving through rock.
CameraOS: answers from the cameras you already own
cameraos.app
CameraOS presents AI video analytics for existing security cameras and recorders, with examples including queue length, free tables, falls, and vehicle activity.
Kavia, the lab-bench voice copilot
kaviaapp.org
Kavia presents a browser-based AI assistant for following lab protocols hands-free, asking troubleshooting questions, and recording experimental steps.
Across the archive
8,065 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.