Grouped by what you are trying to find out. If yours is not here, ask us — a person answers.
You buy credit up front. Every time an agent completes a run, the cost of that run is deducted from your balance. Cost is driven by how much text the run processed, so a short support ticket costs a fraction of a cent and a forty-document research brief costs a few cents. There is no subscription, no seat licence and no minimum spend.
You are not charged. A failed run still appears in your history with the reason it failed, so you can see what happened, but it never touches your balance. We would rather absorb the cost of our own failures than invoice you for them.
No. Credit sits on your account until you spend it. There is no monthly reset and nothing is forfeited if you have a quiet month.
Unused credit is refundable to the original payment method within 60 days of purchase. We do not ask why. Credit you have already spent on completed runs is not refundable.
Every new account gets $5.00 of credit on signup, which is over a thousand Support Triage runs or around a hundred Research Analyst briefs. No card is required to sign up and nothing auto-charges when it runs out.
Because it processes roughly ten times the text. Pricing follows token volume rather than a made-up tier structure, so an agent that reads forty documents costs more than one that reads a single email. The per-run price shown on each agent page is what a typical run costs.
Runs stop. Nothing is queued, nothing overdraws and you are never billed in arrears. Top up and the next run goes through immediately.
No. Each agent has a defined input shape and you fill it in. Where an agent needs to know something about your business, such as your ticket taxonomy or your ideal customer profile, you write that once in plain language and it applies to every run.
Yes, and that is the point. Every agent takes configuration that shapes how it behaves for you: your categories, your tone, your schema, your rules. Two customers running the same agent get meaningfully different behaviour.
Not by default. Agents return structured output that your system applies. That boundary is deliberate: a classification mistake costs you a mis-routed ticket, whereas an action mistake costs you a customer. Where you want an agent to act, you wire the output into your own systems and stay in control of what it is permitted to do.
The accuracy figure on each agent page is measured against our own evaluation sets and is an honest number, not a marketing one. Every agent also returns a confidence score per run, and can be configured to abstain rather than guess below a threshold you set.
Yes. Workflow Runner is built for exactly that: describe a process in plain language and it decides which agents to call in what order, passes data between them, and returns one consolidated result with a per-step audit trail.
Document Extractor and Research Analyst accept PDF, DOCX, HTML, plain text and images through OCR. The others take structured text and JSON. Anything we cannot parse is rejected clearly rather than processed badly.
No. Your inputs and outputs are never used to train any model, ours or anyone else's. This is not a setting you have to find and switch off.
Inputs and outputs are retained for 30 days so you can review run history, then deleted. You can set retention to zero on your account, in which case only metadata such as timestamp, token count and cost is kept for billing.
In the EU and the US depending on the agent. The security page sets out which, and enterprise accounts can pin processing to a single region.
Not routinely. Support staff cannot read run contents. If you ask us to investigate a specific run, you grant access explicitly and that access is logged and time-limited.
Not yet. The observation period is in progress. We would rather tell you that plainly than imply an attestation we do not hold. The security page lists exactly where we stand on every framework.
No. You give an email and a password and you are straight into the dashboard. We ask for an email so we can reach you about billing and so you can recover the account, not as a gate.
Open a ticket from inside your dashboard under Contact Us. Tickets are answered by the people who build the agents, typically within one business day.
Not on the standard plan today. Team accounts with per-member API keys and role separation are the next thing we are building.
Open a ticket and ask. We delete the account and refund any unused credit that is within the 60-day window. There is no retention flow designed to talk you out of it.