Overview & Usage
The first two tabs answer the two questions every App owner asks: is it healthy? and what does it cost?
The health verdict
The banner at the top of Overview grades the App from its last 30 days:
- Healthy (green): error rate at or under 1% and no worrying trend.
- Watch (orange): error rate above 1%, or views dropped more than 20% week-over-week. Nothing's on fire, but something changed.
- Needs attention (red): error rate above 5%. Users are hitting failures.
Next to the verdict, short facts explain why: "up 18% WoW", "2.3% error rate", "5 log errors", "2 tool errors". A quiet month with no problems reads "Clean 30 days". Each fact points at a tab: log errors live in Logs, tool errors in Tools, error rate in API Routes.
Reading the numbers
- Views (30d) with a daily sparkline and the peak day, plus cards for this week, unique users, API calls, and error rate, each with its week-over-week trend.
- Latency (average response time of the App's server routes), storage size if the App has a database, and last active.
- People: who actually uses the App, in three modes. Top ranks by view count, Recent shows who was in last, All is the full list. Counting unique people beats counting views when you're deciding whether an App earned a bigger investment, or retirement.
- Activity summary: tools, agents, and log volume over 30 days, with error counts when they're not zero.
Trends earn their keep before the verdict turns orange. A 30% week-over-week drop in views on a Healthy App often means a team changed workflow, a share got removed, or a link died quietly. The verdict only reacts to errors; the trend tells you about relevance.
Name, description, and icon
The card above the health verdict holds the App's identity: hover the name or description for a pencil to edit them in place, and click the icon to open its menu.
- Suggest with AI designs an icon from the App's name and description. Refine it by conversation (the dialog suggests "use a shopping-cart metaphor, more blue, less busy") and keep the one that fits.
- Upload image (or Replace image once one is set) takes any web image format: SVG, PNG, JPG, WebP, GIF, or ICO. Prefer SVG when you have one; it stays crisp at every size. A file that isn't an image is refused with "That file does not look like an image."
- Clear icon removes it, and the App falls back to the default.
An App keeps exactly one icon, and it serves everywhere the App shows its face: the gallery tile, the browser tab while the App is open, and the home-screen tile when someone adds the App to an iOS home screen. Uploading in a new format replaces the old file. One caveat: an ICO icon still renders in the tab and the gallery, but it can't produce the iOS home-screen tile, so those installs get the standard initials tile instead.
Usage
The Usage tab is the App's AI meter. If the App (or its agents, or its built-in chat) calls Informer's AI, the credits land here:
- This month: total credits with a daily chart, and a month-over-month trend.
- By model: which AI models the spend went to.
- Top users: who's driving the usage. The App's own agents appear here too, marked as agents, so scheduled automation is visible spend, not a mystery line item.
- Resource breakdown: which parts of the App (chat, agents, tools) consumed the credits.
- Build Cost: what the App Builder spent making the App, one row per build, newest first, with each build's model calls and what it cost. Amounts are in credits, or in dollars when the builds ran on your own provider key. The section appears only for an App the App Builder made.
Why you'd use it: when the monthly AI bill moves, this tab tells you whether it was people or automation, and which model tier. An agent on an aggressive schedule shows up immediately as a top "user". Build Cost answers the other question, whether the App cost more to make than it costs to run.
Usage stays visible even without the Apps license, like Overview and Sharing, because billing visibility shouldn't depend on a feature license.