Questions worth asking before the first call.
Plain answers to the things buyers ask us first. If yours isn't here, send it through the form and you'll get a plain answer to that too.
What it costs, and how long it takes.
Nothing. We sit with the person who runs the process, find where it breaks, and demo the build on your own documents, so you see where it struggles as well as where it works. You get a written assessment either way. After that you decide whether to go further.
As one project with a fixed scope. The price depends on the process, the systems it crosses and the exceptions it has to handle. You get it in writing before any work starts.
Six weeks for one process, from mapping to handover. A process that crosses many systems can take longer, and you hear that during discovery, before you commit to anything.
Your team runs it. If you want us to maintain it, that is a separate agreement with its own billing. Nothing stops working if you never sign one.
What the agent does, and what it leaves to people.
Then you already know where it breaks, and that is the part we build for. A bot follows a script and stops on a case it has not seen. An agent works toward a goal, handles the odd case, and hands over what it should not decide.
Before it takes the lead, it runs in parallel with your team and every miss is reviewed each week. After that, anything outside the tolerances you set goes to a person. Every decision is logged with the inputs it read.
Systems with an API, a database you can grant access to, a shared mailbox or a file store. If a system offers none of those, we'll tell you during discovery.
A model from the provider you choose, called through your own account. You hold the terms with that provider and you pay it directly.
It takes the repetitive part of the queue. Decisions about money, people and legal exposure stay with a person. What your team does with the hours is your call.
Who holds what, before and after.
The agent runs in your environment, under your access rules, on your model account. Our access is read-only until you approve more, and it ends at handover. The full detail is on our Security page.
Yes, before you share a single sample. Your template or ours.
You do. The agent, its configuration, the documentation and the decision log all stay with you after handover.
Three senior people who build, and nobody you have to train up on your process. It is also why we take on a few clients at a time.
Didn't see your question?
Ask it in the form. You'll get a straight answer, including when the answer is that an agent is the wrong fix.