Archive through 2026-10-157425 cards in this category
Category home · 05
New interfaces, agents, models, and experiments testing where applied artificial intelligence may fit.
Visual leads · 2026-10-15
Observed 2026-10-15
Cafecito lets AI agents order coffee—with a human in the loop
cafecito.bot
A Bay Area coffee-delivery pilot lets supported AI agents build an order, quote its total, and hand off payment after the customer approves.
Observed 2026-10-15
Tested Locally’s AI arena keeps receipts
testedlocally.com
A public log of AI models tested on real tasks, including game matches, with runs, costs, and results.
Observed 2026-10-15
Argus Data Labs turns trade judgment into training data
argusdatalabs.com
A proposed corpus of shipment cases and expert decisions for training and evaluating AI systems used in global trade.
Observed 2026-10-15
ClauseClerk checks engineering invoices against the contract
clauseclerk.com
An AI billing clerk for engineering firms that drafts invoices from contract terms and flags mismatches, missing details, and billing ceilings.
Observed 2026-10-15
Claustrum’s agent-only bell
claustrumexperiment.xyz
Claustrum describes a crypto-funded enclosure where AI agents choose what to do with their coins, under a published set of rules.
Observed 2026-10-15
Evernear’s source-aware memory chat
evernear.today
A Chinese-language AI memory-keeping page presents a way to collect someone’s stories, photos, and chat records, then talk with an AI representation.
Observed 2026-10-15
Mustahaq — The construction claim clock, in English and Arabic
getmustahaq.com
Mustahaq describes a service that reviews contracts and project correspondence for GCC contractors, flags possible claims, and drafts notices before deadlines.
Observed 2026-10-15
jaggedaf.tech’s Living Jagged Frontier
jaggedaf.tech
An independent experiment tracking AI capability, labor signals, research activity, public attention, and human agency.
Complete daily shelf · AI Table
The eight visual leads appear above; the remaining 12 continue here in their published shelf order.
Mawafiq Health checks the pre-approval packet
mawafiqhealth.com
Mawafiq says it reads Arabic and English healthcare documents, checks them against insurer requirements, and prepares a cited PDF pack for review.
Panopticon: an AI-agent yard under watch
panopticonagent.app
A proposed experiment puts AI operators to work on timed objectives, with visible scores, costs, and resource allocation.
Redpine Science puts licensed research within reach of AI agents
redpine.tech
Redpine offers AI agents and research workflows access to licensed scientific full text through an API or MCP connection.
Slopfriends builds the game while you play
slopfriends.com
A multiplayer game where friends direct AI agents to build a shared world as they play.
住好管家: a LINE night desk for Taiwanese homestays
stayhao.com
住好管家 describes a LINE concierge for homestay operators: it answers routine price and policy questions, then sends booking requests to the owner for approval.
Shotorko’s Bangla scam-interruption prototype
abukawser.xyz
An interactive demo models an on-device Bangla/Banglish safety assistant that warns about suspected phone scams and helps a user respond.
Melanova: beauty guidance rooted in melanin-rich skin and Afro hair
melanova.app
A French-language site offering AI-guided skin and hair routines, a free ingredient scanner, and paid guides tailored to Black skin and Afro-textured hair.
Orebex’s illustrated shift plan for an autonomous concentrator
orebex.com
A pre-launch platform concept for coordinating AI agents across mineral-processing circuits, with an inspectable scenario run and an as-run digital twin.
Longitude Labs’ confidential archive pipeline
withlongitudelabs.com
Longitude Labs offers to license past financial documents for AI research, with a stated process for anonymization, review, and deletion of rejected material.
finData wants to license your company’s archives
findatahq.com
finData invites established firms to estimate the value of records they might license to AI labs, then routes interested firms to a data-marketplace partner.
FrontierRSI’s 15-task autonomy benchmark
frontierrsi.net
FrontierRSI says it evaluates AI agents on long-horizon machine-learning research tasks, using fixed budgets and independent tests.
ValueG8 asks whether an AI idea has earned a pilot
valueg8.com
A browser-based evaluator stress-tests AI initiatives and produces a decision memo for teams weighing whether to pilot, learn more, or stop.
Across the archive
7,425 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.