The Solo Tasks Board: Cogny Now Measures Every Change You Ship

Marketing work has a strange asymmetry: everyone remembers to ship, almost nobody remembers to check.

You pause the bleeding keywords, publish the rewritten landing page, launch the remarketing campaign — and then the next fire starts, and nobody ever goes back to ask did that actually work? The loop stays open. The learning evaporates.

Cogny Solo now closes it. As of today, every change you ship from the chat lands on a Tasks board, and Cogny comes back on its own to measure whether it worked.

How it works

If you've used Cogny Solo lately you know the front half of the loop: run a skill (an SEO audit, a Google Ads audit, a Meta audit), get a report that ends with a concrete next step, and let the agent implement that step for you — it connects to the end system (Google Ads, Meta, your CMS, GTM) and ships the change itself.

The new part is what happens after:

1. Shipped. The moment the agent creates something real — a campaign, an ad set, a blog post, a GTM container publish, a set of negative keywords — it appears as a card on the Tasks board (new in the top menu). No manual logging.
2. Analyzing. Each task gets a scheduled impact check, timed to the kind of change: ads and tag changes are checked after about 3 days, content and site changes after about 7, since organic effects take longer to show. When the check comes due, Cogny re-reads your connected data sources.
3. Impact measured. The analysis compares equal-length windows before and after the shipped date, attributes conservatively (the campaign's own metrics, the page's own traffic — not account-wide noise), and writes a plain-language verdict on the card: it worked, partial win, didn't move the needle, or inconclusive — with an estimated dollar impact when the data supports one. If the change is too young to judge, it says so and checks again later.

The board is deliberately not another to-do list. You can't add tasks to it. It only contains things that actually shipped — which means everything on it is real, and everything evaluated on it is evidence.

Share the wins

Every evaluated task gets a Share button that renders a result card — what shipped, on which platform, what the measured impact was — that you can copy as an image, text, or link and post wherever you like. When your remarketing campaign turns out to have paid for itself in a week, you shouldn't have to screenshot a dashboard to prove it.

The link is personal: every user who signs up through it and becomes a paying subscriber earns you 200 credits, and they get their first month of Cogny Solo free with code FREEMONTH. Bragging pays, literally.

What it costs

Impact analyses run on your workspace's credit balance, the same top-up credits that power chat. Each check is a short, bounded agent run — a handful of read queries and a verdict. If your balance hits zero, tasks wait in a "top up to evaluate" state and resume when you top up. Nothing is ever analyzed twice behind your back: one verdict per task, plus automatic re-checks only when the first look said "too early to tell."

Why this matters

The point of an AI growth agent isn't to produce more output — it's to compound. Compounding requires knowing what worked, and knowing what worked requires someone actually going back to measure. That's the step humans skip, reliably, forever, and it's exactly the kind of unglamorous scheduled diligence an agent is good at.

Open your Tasks board, ship something from chat, and let Cogny tell you whether it was a good idea.