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 2026-10-167442 cards in this category
Category home · 05
New interfaces, agents, models, and experiments testing where applied artificial intelligence may fit.
Visual leads · 2026-10-16
Observed 2026-10-16
RSI Watch: the recursive self-improvement claims ledger
rsi.watch
A dated, sourced tracker of public claims about whether AI has reached recursive self-improvement.
Observed 2026-10-16
agentscanr: AI agents put through UK business tests
agentscanr.com
A UK-focused AI-agent testing site publishes benchmark results, costs and run logs, alongside a browser-based prompt checker and paid testing services.
Observed 2026-10-16
Practice AI Review’s dental-AI scorecard
practiceaireview.com
A public directory and comparison rubric for AI tools aimed at dental practices, with dated scores, pricing details, and vendor-stated BAA status.
Observed 2026-10-16
Apex-3 trains on one GH200—in public
apexsi.cloud
ApexSI presents language models alongside their weights, training code, logs, and research.
Observed 2026-10-16
Saindra, an open-canon AI club singer
saindra.gallery
A fictional AI-made Latin club singer, presented through music videos, a character profile, visual looks, and an invitation to reuse her under CC BY 4.0.
Observed 2026-10-16
cato8: a private AI computer with a physical network switch
cato8.com
cato8 presents a local-AI computer with models pre-installed, setup by an engineer, and a physical switch intended to disconnect it from the network.
Observed 2026-10-16
EVES’s 432-question answer check
edenverification.com
A public evaluation page for Eden’s AI verification system, comparing how often it made up answers with three models answering from memory.
Observed 2026-10-16
Harpia shows its WhatsApp support workflow
harpia.health
A detailed account of an AI system drafting and answering WhatsApp support messages for companies in a health-services group.
Complete daily shelf · AI Table
The eight visual leads appear above; the remaining 9 continue here in their published shelf order.
Command Intent gives an RTS a radio net
metag.studio
An in-development real-time tactics game proposes voice-commanding a platoon whose AI soldiers answer, ask questions, and carry out orders.
StatAgent’s AI workflows for biostatistics
statagent.org
A free collection of copy-paste workflows for using AI assistants on biostatistics tasks, including drafting SAP sections, translating SAS to R, and checking tables.
The Peaceful AI
thepeacefulai.com
A chat AI project says it uses Jainism’s five vows to guide its answers and refuse requests that would help harm living beings.
AcehX: language technology for Acehnese
acehx.com
AcehX documents language-technology work for Acehnese, including spelling correction, translation design, research, and datasets.
Avestiq’s proposed ionic-liquid model
avestiqx.com
Avestiq describes LILM, a physics-informed model in development for ranking ionic-liquid candidates against industrial performance, safety, life-cycle, and cost criteria.
inv: No Monkeys Required, 16 GB of RAM
nomonkeysrequired.com
A downloadable AI assistant for macOS that says it runs on your computer without an account or cloud connection.
AI rankings built from work, not benchmarks
usecaserank.com
Use Case Rank collects AI task runs and ranks models using recorded costs, completion times, and human interventions.
Lilly Intelligence turns stock questions into quotes
lillyintelligence.com
A Namibia-built AI sales agent presented for businesses that sell from stock and quote prices, with customer conversations gathered across messaging channels.
Cresyte: an AI test bench for power boards
cresyte.com
Cresyte is developing an AI agent that turns a board datasheet into a test plan, operates bench instruments, and produces a report.
Across the archive
7,442 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.