The two fears in every AI purchase: per-seat pricing that scales with headcount instead of value, and a usage bill nobody can forecast. Our answer is a flat workspace price and a clear pool of completed runs, with overage you can compute on a napkin and hard budget caps in the product itself, so the ceiling is something you set, not something you discover. You pay for completed runs, never seats or tokens.
₹0forever
+ 18% GST
100 runs included · then ₹20.00/run
₹3,499per organization / month
+ 18% GST
300 runs included · ₹11.663 per completed run · then ₹15.00/run
₹19,499per organization / month
+ 18% GST
2,500 runs included · ₹7.800 per completed run · then ₹12.00/run
from $2,000annual agreement
A run is one completed piece of work, however many agents it takes. INR prices are exclusive of GST.
Team is $299 a month and includes 5 AI employees and 2,500 completed runs. Beyond that, runs are $0.15 each from prepaid credit — you top up deliberately, so there is no invoice you did not agree to. That is the whole formula, and failed runs never count.
The bill you can't forecast is the one that kills the renewal. So cost control is a product feature, not a billing apology: every message carries its own token and dollar cost, budgets are hard caps, and breakers pause a runaway worker before it burns a quarter's budget.
One completed execution of an agent, team or employee task, trigger to finish, however many internal steps it takes. A five-step Sequential Agent finishing once is one completed run, not five, and an AI employee that calls three agents and searches a knowledge base to finish one job is still one run. Every run shows up in Analytics with its own trace and cost, and the internal fan-out is shown but never charged.
You are not billed for it. A run only counts when it completes its task. Runs that error out, time out, or that you stop mid-flight never touch your pool. You pay for outcomes, not attempts, so a flaky integration or a worker you paused is not something you find on the invoice.
Yes. Unused runs from your monthly pool roll over for one month, then expire. We'd rather you keep what you paid for than watch a meter reset at midnight.
Work already running finishes; only new work pauses. You can either move up a plan or add pay-as-you-go credit and carry on where you are. Credit is prepaid, so there is no invoice you did not agree to, and it does not expire.
You bring your own provider keys, on your own agreement with OpenAI, Anthropic, Google or whoever else, and pay them directly at their rates. Model usage never passes through our billing, so there is no margin on tokens and nothing hidden inside a credit. Every message still records its own token count and dollar cost in Analytics, so you can see exactly what a run cost you even though we did not charge for it.
The product, not just the invoice. Every message records its token and dollar cost, budgets are hard daily and monthly caps with a forecast breach date, and circuit breakers auto-pause a worker when a cost or error threshold trips. A runaway loop hits a ceiling you set, not a bill you discover.
Not yet. The application and its workers run in a single region today (Azure East US), with the self-hosted integration gateway in Azure Central India, and self-hosted and dedicated deployment are on the roadmap rather than shipping. If either is a hard requirement for you, talk to sales and we'll be straight with you about timelines.
Packs are free to install on any plan. The systems they stand up consume runs from your pool like anything else you build. No per-pack fees, no surprise line items.
A working system in fifteen minutes, on the free tier. No card, no countdown.
We use analytics cookies to see which pages help and which don’t. Nothing loads until you choose. Cookie Policy
Hello there.
AI agent. It can make mistakes, and a human reviews anything that matters.