0.1.21: AI Overview tracker, Facebook Group posts, LinkedIn post engagers

0.1.21 adds three new sources to the desktop app, lets you describe a Detail scraper column in plain words, and lets you sign in without a Google account.

Three new sources

• AI Overview tracker. New table → Research. It runs each of your searches several times in a fresh, logged-out browser and counts what came back: which products Google's AI Overview names, which pages it cites, and whether your own domain is cited. You get one row per search per day, with a line saying what changed since the last check. It also opens the most-cited pages and measures what they have in common (length, tables, lists, FAQ). Every tracker starts with a ready "What Google rewards" AI column that you can run on your own model.
• Facebook Group posts. New table → Social. Paste one or more group addresses and every post becomes a row with its author, text, time, link, and its likes, comments and shares. Later checks add the new posts and refresh the counts on recent ones. It runs in the built-in browser on your own Facebook login. Public groups need no membership; a private group needs you to be a member. A new post costs one action; refreshing counts is free.
• LinkedIn post engagers. New table → Social. Paste LinkedIn post links and get the people who commented on or reacted to them: profile, headline, job title, company and their comment. It reads with your own LinkedIn login in the built-in browser, at a human pace.

The tracker and the Facebook source can run on a schedule: turn on Auto-refresh in the table's Automations panel.

Scrapers

• Describe a column. In the Detail scraper, say what you want from each row's page ("the article text, without the related links") and your AI model writes the column once. After that, rows are read without AI. Refine rewrites one column when it is not quite right.
• News and blog lists. The List scraper now recognises cards that are themselves links, so a page of article cards comes back as headline, link, date and image.
• Readable article text. Text keeps its paragraphs and line breaks instead of running together, and a column can keep its links.

Sign in with an email code

You no longer need a Google account. Enter your email, and type the code we send you.

Fixes

• Auto-run stops retrying a failing row. A row that kept failing was retried every minute. Retries are now capped, and a column you stop stays stopped.
• HTTP request columns handle quotes and line breaks. A value containing a quote or a line break used to fail the row with "Body is not valid JSON". Values are now escaped.
• "Updated" times stay put. Cloud sync was re-stamping every table as updated on each sync. A table's time now changes only when the table does.
• Record details and the Sent log are tidier and easier to scan.

Get it

Already installed? Manycrawl updates itself, so 0.1.21 arrives the next time the app starts. New here: download for Mac or Windows.