Real jobs for a team of your AI apps, each shown from a real run.

Two teams, two real runs. See the team that does each job, and what LoomWatch does to make it work. The first is told word for word from its run; the second ran on its own schedule. The screenshots are of those same runs, in LoomWatch.

01 A real run

Learn anything from first principles.

Ask about anything. Mapper digs it down to what you already know, and you say where to start. Teacher builds it back up one idea at a time. Then Checker, on another company's app, checks every step. All three read the same skill: explain like I'm 5.

Mapper Claude→You→Teacher Claude→Checker Codex

Learn from first principlesa recorded runRecorded

Read the whole lesson, as the team handed it back

02 A real run, every morning

A daily tech digest, checked against sources you trust.

Every morning at 9:00, one agent gathers the last day's stories from the sources you listed. A second reads each story at its sources and keeps only what they stand behind, and a third writes it up. Nothing waits for you: the digest is done before you open it. Scheduled runs happen only while LoomWatch is running and your Mac is awake.

Every day 09:00→Gatherer Codex→Reader Claude→Writer Claudeall three read sources.md

Your sources, in a file: The trusted list and the rule for keeping a story sit in the team's Brief, a Markdown file every agent reads. Edit it, and tomorrow's run follows it.
Two sources, or it's out: A story stays only if its own primary source says it, or two independent newsrooms report it. Reader opens every source it can and checks the rest through web search, never on the headline alone.
It runs on its own: No stop for you. In a scheduled run anything not switched on is declined, so only web search is on, and only for Gatherer and Reader.
Two companies at one desk: Codex gathers, Claude reads and writes, so one model never decides alone what's true.
Teach it once: Write "fewer funding stories" in the Notebook and keep it. Every later digest reads it.

The request it runs

Today's tech news digest for {{weekday}} {{date}}: what happened in the last 24 hours, confirmed by the trusted sources in the Brief.

That run, by the numbers

12stories gathered from the list
7kept by Reader under the rule
7in the digest, in 8 min 48 s

Seven of the trusted newsrooms won't let Claude open their pages. The Brief names them, so Reader checks their stories through web search instead. The receipt shows every step finished, with nothing flagged.

What comes back example

AI

A lab ships a smaller model that runs on a laptop. The company's release notes give the sizes; two newsrooms confirm the benchmark claims. release notes newsroom newsroom

Security

A widely used library patches a remote-code flaw. The maintainers' advisory names the affected versions. advisory newsroom

daily-tech-digest.yamlthe team file

daily-tech-digest.brief / sources.mdthe trusted sources and the rule, in its Brief

∴ Coverage

What each team leans on.

Each mark is a line in that team's file, not a promise. Pick a column to open that use case.

What LoomWatch does

Your job isn't here? Describe it, and Ask LoomWatch proposes the team.

Get LoomWatch