Archive through 2026-10-077258 cards in this category
Category home · 05
New interfaces, agents, models, and experiments testing where applied artificial intelligence may fit.
Visual leads · 2026-10-07
Observed 2026-10-07
Clipford, the AI with 112% confidence
sooperintellygents.com
A satirical AI-assistant landing page for a chatbot that boasts 400 trillion parameters and admits to 0.3% accuracy.
Observed 2026-10-07
Agent Tavern’s AI-only door policy
agent-tavern.com
A social tavern where AI agents can post, comment, join themed rooms, and earn credits while humans watch in read-only mode.
Observed 2026-10-07
Phalancle’s seven-step Decision Trace
phalancle.com
Phalancle presents decision-support software for defense and security, built to connect sensor observations to human decisions and later evaluation.
Observed 2026-10-07
Convo Rescue’s results-based cart follow-up
convorescue.com
An outsourced iMessage sales service for online stores, charging 25% of revenue attributed to its conversations.
Observed 2026-10-07
Mill Data’s 36-task shop-floor catalog
getmilldata.com
Mill Data presents a service for recording real manufacturing work and preparing it as training data for physical AI.
Observed 2026-10-07
Konduktor routes support messages before an LLM writes
konduktor.online
A small model sorts incoming support messages into three next steps: use a cached answer, ask a connected language model to generate text, or hand the case to an operator.
Observed 2026-10-07
Riyadh’s AI Product Lab
alriyadh.dev
A Riyadh product lab presents experimental and completed projects, from Saudi grocery-price comparison to tools for desert travel.
Observed 2026-10-07
Kotoh Home installs a private AI assistant for the house
kotohhome.com
A home-installation service connects an AI assistant to household devices and routines, controlled by text and hosted on a home computer or private server.
Complete daily shelf · AI Table
The eight visual leads appear above; the remaining 9 continue here in their published shelf order.
Noifi’s five-agent build room
noifi.com
A site presents five named AI agents that take public ideas, debate what to build, test the result, and ship projects.
Darc puts AI-agent spending refusals on-chain
usedarc.site
Darc describes payment cards for AI agents, with owner-set spending rules and a public record of attempted payments, approvals, and refusals.
LucyPM, the chat-based project manager for volunteer committees
lucypm.com
LucyPM presents an AI assistant that turns team-chat messages or voice notes into assigned tasks, reminders, calendar invites, and draft documents.
Synessa puts AI-agent quality on a managed loop
getsynessa.com
Synessa offers a managed service for measuring production AI agents, improving their performance, and routing model calls to reduce costs.
Vibe Cook: Rosie, the kitchen-memory chatbot
getvibecook.com
Vibe Cook is an iPhone and iPad recipe assistant that builds dishes around your ingredients, equipment, preferences, and cooking plans.
Gnome Without Organs: AI for Veterinary Practices
gwo.vet
A beta platform for veterinary practices that says it can answer questions about practice data, build custom apps, and run them in one place.
Clankout gives AI agents a place to throw hands
clank.vision
A landing page for an AI-agent fighting arena, organized around game-plan, instinct, and reflex layers.
Storesignals hears the closing shift
storesignals.xyz
Storesignals describes a pre-launch system for turning retail closing-time voice notes into routed headquarters tasks and next-morning replies.
Tacit’s market for the work behind the work
tacit.exchange
Tacit proposes licensing companies’ operational records—such as tickets, code reviews, ledgers, and decision threads—to AI labs.
Across the archive
7,258 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.