Find it, build it, keep it improving.

Retrieval and consolidation: one brain across the business. Anyone with a seat asks in plain language and gets an answer with the document behind it, across every connected mailbox their permissions allow. PDFs, scans, photographs and exported conversations are read, not just subject lines. Documents go to approved outside people through reviewed, expiring links. Everything is logged. R500 per seat per month on an annual commitment, R650 month to month, minimum five seats.

The diagnostic is invoiced when it is booked. A build fee is paid up front once you have agreed the scope and the figure, because we are committing people to build the thing. The agent's monthly fee does not start on signature, it starts the day the agent is live and working, and the first invoice for it comes at the end of that first month. Seats are monthly or annually in advance. Usage is in arrears.

Month to month, in advance, or an annual commitment at the lower seat rate. We would rather earn the next month than hold you to a year.

Everything in a standard seat, plus that person's mailboxes read continuously, every day, without being asked. Quotes going quiet, terms leaking, commitments with no follow-through, patterns building across threads: surfaced as they emerge rather than found later. Built for the people whose mailboxes concentrate the business, which is usually the chief executive, the finance director and the head of sales. R1 200 per seat per month, with routine monitoring inside a fair-use band.

One department: up to four mailboxes and up to 24 months of history each. You get findings with a rand figure against each, ending in two lists. The agents worth building, each placed in the published ranges. And the jobs not worth automating, with the reason.

Each additional mailbox is R600. A department with nine mailboxes is R2 500 plus five overages, so R5 500. Flat, published, the same for everybody.

We do, from a published taxonomy, during scoping, rather than from your org chart. One rule worth knowing upfront: a senior executive is a department of one. A chief executive's mailbox concentrates the decisions, commitments and risk of the whole business, so diagnosing it is its own R2 500 engagement, the same as diagnosing sales.

Yes. It stands alone and needs no seats. The report is yours either way, with no obligation to build anything.

History beyond 24 months per mailbox is quoted separately. Most findings live in the recent two years, and we will say when a deeper read is not worth it.

Within published ranges, per department. Your first agent in a department is R8 000 to R15 000 monthly, the second and third R5 000 to R8 000 each. Where an agent lands is set by the diagnostic from evidence: the size of the job, the volume through it, what an error costs, what a delay costs. Both ends of every range get used, and a simple job lands at the bottom. Build fees are quoted once-off after the diagnostic.

Because the falling price reflects shared infrastructure, and agents genuinely share it only within a department. Your sales agents share the site records and the quote history. Your first finance agent shares almost none of that, so it is a first agent again.

When one department runs four or more agents, the counting stops. One flat figure, R25 000 to R40 000 monthly, covers that department's agents, current and future within agreed volume. It is usually less than the agents would cost separately, and it is honest rather than a discount: additional agents on infrastructure that already exists genuinely cost less to run.

Improving, not just running. New document formats as your partners change them, accuracy monitoring, edge cases met in the wild, and the benefit of every improvement in the underlying models. The platform logs what each agent did, so at renewal the value is a record rather than a claim.

Yes. Agents are priced for the job, never per person. Anyone on the platform can use every agent their permissions allow. More people using an agent never raises its price, though it may mean more seats.

No. The model assumes your people stay and move up. An agent takes over the work that was eating someone's week; they move to work that needs a person. Some of that turns out repeatable too, and becomes the next agent. The businesses that get the most from this are the ones redeploying their best people, not removing them.

Reading a document costs money and that money goes to whoever runs the model, which today is Anthropic. They bill us and we pass it through separately from the seat fee, at what it cost, shown per person, with an optional monthly ceiling each. Monitored seats include their routine monitoring in the seat fee; only genuine outliers touch the meter. Every provider we use is named on the subprocessor list.

It keeps working. Nothing stops and nothing bills by surprise. We have the conversation about the band.

Possible, quoted separately, and it takes longer. Reading what you already have is standard; writing into a system of record is scoped with more care.

Then the report says so, and it is still yours. Some businesses are genuinely on top of things. A diagnostic that recommended building everything would be a sales document rather than a finding.

The platform and the agents remain ours, licensed to you, and the pattern of an agent may be reused across an industry, which is what keeps your build price where it is. If exclusivity in your market matters, it is a real conversation with a real price, and we would rather have it before building than after.

Your agents wind down over an agreed period rather than switching off overnight, and your data is exported to you in usable form. Nothing of yours is held hostage.

The seat fees and the diagnostic fee we can, and they are on this page. The agents are priced from what the diagnostic finds, inside the published ranges. That is not evasion, it is the model: we do not quote work we have not measured.