Tally, the pricing Grok Bot
The pricing model, then the pricing page
Ramp, the onboarding Grok Bot
First run, activation and empty states
Lift, the conversion Grok Bot
Ranks the changes that move the number
Radar, the competitor tracking Grok Bot
What changed this week, with URLs
Track, the analytics Grok Bot
The events you would actually query next month
Risk, the churn Grok Bot
Who is leaving and the one action this week
Grok Bots for the numbers: working out the pricing model before the pricing page, designing onboarding around time to first value, ranking conversion changes by what they are worth, planning the analytics events you will actually query, and finding the accounts that are leaving before they say so.
Growth work fails in a predictable order, and these are built to work in that order rather than to reach for the nearest lever. Tally decides the pricing model before the pricing page, because a page cannot rescue the wrong model. Lift works down from the offer and the message match before it ever reaches a button. Ramp asks what the first real thing in your product is before designing anything around it.
Two of them run on a schedule, which is what makes them different from an analysis you commission once. Radar checks competitor pricing, changelogs and careers pages weekly and reports only what moved. Risk scores accounts every Monday and returns the ones leaving, ranked by revenue at risk, with one action each. Churn is a lagging number and the signals are not.
Tally for when you are about to build a pricing page, or the revenue does not grow with the account
Ramp for when signups are healthy and activation is not
Lift for when the page exists, the traffic arrives and the number is flat
Track for when you are about to build a feature and will want to measure it
Risk for when you want to know who is leaving before they say so
Radar for when you keep finding out about competitor changes from customers
Because the page cannot fix the wrong model. Per seat punishes the thing you usually want, which is more of the company using the product. Tally decides what to charge for, then builds the page around it.
Time to the first real thing. Not account created and not tour completed, but the moment the product does what somebody came for. Everything else in a first run is measured against that.
No, and it is told not to. It works down from the offer and the message match, through the clarity of the first screen and the objection nobody answered, and only then reaches craft. It says when the problem is upstream of the page.
Yes. Radar and Risk carry routines, so Grok Bot checks the pages or scores the accounts on a set day and reports what changed without being asked.
Pricing pages, onboarding flows and the screens these Bots change all get built somewhere, and there are design templates for each tool: Lovable templates, Bolt templates, Replit templates, Claude Code templates, ChatGPT Codex templates, Cursor templates, Windsurf templates, Framer templates, v0 templates and Cline templates.