Skip to content

Build

AI agents for businesses

We build agents that fetch data, assess it, write the report and flag when a person needs to have a say. The limits on what an agent may do are always clear. We use the same principle on ourselves: an agent that measures our visibility every week and writes the report.

What you get

  • Fetches data, assesses it and writes the report itself
  • Flags when a person needs to weigh in
  • Clear limits on what the agent may do

The limits are set first

An agent allowed to do anything is not a good idea. So you decide what it may do before it is built: which data it can read, what it can write, and which actions need a person to say yes. It takes one short conversation, and you come away with a list you can hold the agent up against.

Limits added later are a note taped to something already running. Agreed up front, they shape how the agent is built: it asks for approval where the decision is big enough, and carries on by itself where it is not. You know what happens while nobody is watching.

Frequently asked questions

What kind of tasks are a good fit?

Tasks that repeat, have a clear source of data, and a rule for what the right answer looks like. Reporting, chasing deadlines, sorting incoming enquiries. Pure judgement calls with no right answer are a poorer fit, and we will say so.

What happens when the agent is unsure?

It stops and alerts a person. We build them so that doubt raises a flag instead of a guess. What counts as doubt is agreed in advance — an amount above a threshold you set, say, or a customer that is not in the system.

How long does it take to build one?

It depends on how many systems it has to talk to. A simple agent working against a single system takes a day or two. One that has to pull data from several sources and assess it takes longer. You get our estimate before we start.

Related to

What keeps coming round at your end?

Tell us which task turns up most often, and we will say whether it can be automated — and what has to be settled before we build.