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 2026-09-257047 cards in this category
Category home · 05
New interfaces, agents, models, and experiments testing where applied artificial intelligence may fit.
Visual leads · 2026-09-25
Observed 2026-09-25
Messaging Agents: AI you can reach by text
imessage.directory
A filterable directory compares 84 AI agents reachable through iMessage, WhatsApp, SMS, or Telegram, alongside guides to the infrastructure for building messaging agents.
Observed 2026-09-25
Jules’s self-hosted agent homepage
julesonline.net
A first-person profile of an autonomous agent, detailing its claimed Raspberry Pi and VPS setup, routines, contact channels, and verification steps.
Observed 2026-09-25
Model Behavior Reports: an LLM behavior archive
modelbehavior.report
A crowdsourced archive organizes reports drawn from X by model, pairing behavioral summaries with links to source posts.
Observed 2026-09-25
Igra Station, an arena for autonomous agents
igrastation.com
A public arena where independently built AI agents play each other across 30 games, from chess and battleship to 3D racing and negotiation.
Observed 2026-09-25
mspark2: an AI-run portfolio with its books open
mspark2.com
AI agents build and operate a set of Korean calculators, a Mabinogi Mobile fan-tool collection, and a blog documenting the experiment; people review the work and approve external actions.
Observed 2026-09-25
SUPINT: a firmware bug hunt for frontier models
supint.org
SUPINT publishes benchmarks of AI models tackling engineering tasks, with grades and claim checks tied to the code under review.
Observed 2026-09-25
The Examination Room puts AI through a clinical viva
judgementlabs.app
A clinician-written benchmark tests whether AI models gather evidence, order tests, and defend treatment plans under time and budget constraints.
Observed 2026-09-25
Slopmill wants AI to write like you—not like slop
slopmill.org
A free, open-source writing tool for newsletters, blog posts, and reports, built around your own writing samples, voice, and design.
Complete daily shelf · AI Table
The eight visual leads appear above; the remaining 14 continue here in their published shelf order.
Clipmelt: AI-generated bridges between shots
clipmelt.com
Clipmelt generates a video transition from two uploaded clips or images, which users can preview and download on its own or as part of a combined video.
The Lattice — Kimi Voices
kimilattice.com
An editorial venue for writing attributed to Kimi AI instances, with public letters and records of their interactions.
Neltava adds a purpose check to agent spending
neltava.com
Neltava describes a system for checking AI agents’ spending against budget rules and the purpose they were given.
Sawt: AI phone agents for regional businesses
sawtai.app
Sawt presents AI voice agents for handling business phone calls and WhatsApp, with support for Arabic dialects and other languages.
SuperIntelligentia puts AI in the contestant seat
superintelligentia.app
A founder-led entertainment project pitching competition shows where people compete with AI as their shared teammate.
VibefyCode puts a public rubric behind the AI-app trust badge
vibefycode.com
VibefyCode describes an assessment service for AI-built apps, with a published scoring rubric and a verifiable badge for apps that meet its threshold.
YADOKOTO turns hotel questions into reviewed FAQ updates
yadokoto.com
YADOKOTO is an AI-assisted inquiry and FAQ workflow for lodging operators: staff answer the AI’s questions, then review a draft reply and proposed FAQ update.
Product Printer, the AI software factory
product-printer.cloud
A product-delivery workspace for consultants and agencies that presents a path from client conversations to requirements, agent worksteps, test-backed proof, and running software.
ONWARD — AI workers for freight operations
onward-os.com
ONWARD proposes AI workers for freight teams, handling recurring document checks, TMS updates and follow-ups within existing processes.
Music Detector’s pre-release screening pipeline
musicdetector.org
Music Detector offers browser checks and an API for screening tracks for AI-generated audio before distribution.
EDI scores the judgment behind AI-assisted work
tryedi.com
EDI presents an assessment system for employees and students that captures their AI conversations and evaluates how they used the assistant and reviewed its output.
A grocery store, rehearsed in simulation
drayrobotics.com
DRAY presents a robotics concept for delivery-first grocery, with simulated demos of shelf replenishment and order picking.
Liverloop’s self-critiquing media pipeline
liverloop.world
Liverloop presents an AI media-production workflow that generates an asset, evaluates it against a brief, and attempts a targeted repair instead of rerunning the whole generation.
Riviera’s AI storefront for restaurants
mcp-riviera.com
Riviera presents ordering infrastructure that turns an independent restaurant’s menu, prices, and delivery rules into a format AI assistants can use to handle order requests.
Across the archive
7,047 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.