Keep costs predictable with rate limiting
The fear with usage-based AI is the surprise bill. Stark Chat puts you in control: cap what each question can cost and how many can run per hour, on every knowledge base — so usage can never run past what you have decided to spend.
Limits you set, per knowledge base
Every knowledge base has two dials: how rich a single query can be, and how many queries are allowed per hour. Query depth is a simple multiplier — a 2x query allows twice the tokens and consumes two queries from your allowance — so capping the multiplier caps what one question can cost. Set both once in the dashboard and the knowledge base enforces them for you.
Different knowledge bases can have different limits — generous for your internal team, tighter for anything public-facing.
Make public knowledge bases safe to run
A public knowledge base means anyone can ask it questions — which is the point, but also the risk. Rate limits cap what any visitor can consume, so a burst of traffic or a bad actor cannot run up your bill.
That makes it practical to put a knowledge base on your website for customers and prospects, not just behind a login for staff.
Predictable spend, no surprises
With per-query and per-hour ceilings in place, your worst-case spend is a number you chose, not one you discover on the invoice.
Adjust the limits any time as usage grows — no redeploy, no support ticket, just a change in the dashboard.
Set your spending caps in the dashboard and let the knowledge base enforce them — costs stay exactly where you put them.
Try for freeSpend what you decide, nothing more
Cap what each question can cost and how many run per hour on every knowledge base — usage can never run past the limits you set.