Launching today
Throngwatch
Read your Discord without reading it.
11 followers
Read your Discord without reading it.
11 followers
One Standing Brief for your Discord — a 0-100 Health score, a verdict, and top actions, each cited to the message. Free analytics; pay-as-you-go AI.
How does the pay-as-you-go AI pricing actually break down per brief, and is there a cap if a server suddenly gets way more activity than usual?
@polatkeptfc Good question! The math: 1 credit = 1 full AI scan of up to 1,000 messages (ceil-rounded 2,500 messages in a week = 3 credits), and a generated Standing Brief report is a flat 1 credit on top. New workspaces get 3 credits free at signup; extra credits are $2/credit on the free tier, down to $1.20–$1.50/credit on paid plans.
On the spike question no surprise bill. It's a prepaid balance, not metered overage: if a server suddenly gets a lot more active and you run out of credits mid-scan, it just pauses and asks you to top up rather than auto-charging you for going over. Additionally u can set a hard-cap for msgs processed per channel, for example 100 or 2000 per scan, so it is pretty manageble even on early stages.
The cited-to-message breakdown is a really thoughtful touch - so much of Discord moderation is just guessing where the chaos started. Love that the free tier gives you the score without locking the actual context behind a paywall.
@merjaey Thanks! That was the whole point, a health score nobody can verify is just noise, so every action links straight to the message that triggered it. When I was using similar tools, they'd gate basic stats behind a paywall too. I decided to only bill for the parts that actually burn tokens and analytics stay free, AI insights are pay-as-you-go.
The cited actions are genuinely useful, showing exactly which message triggered each flag instead of just dumping a score. Surprised how clean the verdict panel is for something this new.
@guvercinme55192 Appreciate that! The verdict panel went through a few redesign passes - first version was much more "dashboard-y" with more numbers on screen. Realized the score + verdict + citations is basically all that matters, so cutting the noise around it took longer than building the analysis itself.
The cited messages feature is honestly the standout for me. Makes it easy to verify what the AI flagged without scrolling back through hours of chatter. Wish the free tier went a bit deeper on history.
@zlem78230489629 Totally fair, that's the real constraint on the free tier. Deeper history means running the analysis over a lot more messages, which burns real AI tokens, and I can't subsidize that without sponsors. One idea I'm kicking around: open-sourcing part of the analysis pipeline so people who want more history could run it themselves with their own model/API key instead of paying me directly. Not committed to it yet, but it's on the table.
The cited messages tied to each action saved me from digging through threads to verify what was actually said. The 0-100 score is blunt in a useful way, and the free tier is generous enough to actually evaluate a community before paying anything.
@sevimtt5u Really glad that came through - didn't want the free tier to be a teaser you can't actually evaluate on.