home/how we work

Small steps, each one paid back before the next.

You never buy a transformation. You buy one working thing, see what it saves, then decide. That is the whole method; the rest is detail, and the detail is below.

Four steps. You can stop after any of them.

day 0

A 30-minute call

You tell us where the hours go. Within two days you get a short written note: the two or three jobs AI should take first, and what each would take to build.

weeks 1-2

A working prototype on your real data

Not a slide deck. One job, built for real, running on a sample of your actual inbox, calls or documents, so you decide on evidence.

weeks 3-8

Into your tools, tested, handed over

The prototype becomes a system inside the tools you already use, with the exceptions routed to a person, documented, and your team trained to run it.

ongoing

Kept working, measured, extended

We keep it running, count the hours it gives back every month, and add the next job when your team finds it.

What each step gives you.

step 1 · quick-check · free

A written note

  • The two or three jobs to start with, with the hours each takes today
  • What the agent would do, and the guardrails
  • A fixed price for the two-week prototype
30 minutes
step 2 · sprint · fixed price

A working prototype

  • Running on a sample of your real data
  • Your team clicks through it and judges
  • A decision: build, change, or stop
two weeks
step 3 · build · fixed price

A system in production

  • Inside the tools you already use
  • Documented, tested, the team trained
  • Source code and data handed over; they are yours
three to eight weeks
step 4 · run · monthly

Hours counted, every month

  • Monitoring, updates, a person to call
  • A monthly count of what the agent did and saved
  • The next job added when your team finds it
ongoing · cancel monthly

Fixed prices. No surprises. No numbers on this page.

Every step has a fixed price agreed before it starts, so you always know what the next decision costs. We do not publish a price list because the right scope depends on your volume, your tools and your rules, and guessing it from a web page would be a disservice. The quick-check ends with a written proposal and the exact figure for the next step.

What we need from you, and what you get from us.

from you

  • A person who knows the weekOften the office manager or whoever does the admin, for the call and fifteen minutes a day during the prototype.
  • A real sample of the workA week of emails, a day of calls, ten documents. Anonymised if needed. Invented data proves nothing.
  • Your rules, written down onceWhat you promise, what you never promise, who signs what. We write them with you; the agent applies them every time.
  • A decision at each stepBuild, change, or stop. All three are fine.

from us

  • One person accountableThe founder of bss works is in every project, from the first call to the system running. You never explain your business twice.
  • Working software, not slidesEvery step ends with something running: a prototype, a system, a monthly count.
  • Code and data you ownEverything we build for you is yours, documented, portable. No lock-in, by design.
  • A straight answerIncluding “AI is not your problem”, when it is not. We would rather say it in thirty minutes than sell three months.

Rules we do not bend.

Read-only first

Every agent starts by watching and drafting. It only sends, books or files on its own once you have seen it be right, step by step.

Your data stays yours

Agents run inside your own accounts and tools. Where a model is needed we use EU-hosted options by default, and nothing is used to train anyone’s model.

A person for the exceptions

Anything the agent is unsure about goes to a named person, with the facts already gathered. You decide what it may do alone, and when.

Where and in which language.

Founded in Monaco. Working with companies across Europe and the United States, on site along the Riviera or remotely anywhere, in English and French. The agents answer your clients in their language.

The questions that follow.

What happens if the prototype does not convince us?

You stop, keep the prototype and everything we learned, and pay only for the sprint. That outcome is rare, because the quick-check filters the jobs that would not pay back, but it is designed in.

Can we run it ourselves after the build?

Yes. The code is yours and documented, and we hand it to your IT if you prefer. Most clients keep us for the run step because the monthly count and the next-job pipeline are where the value compounds.

How many projects at a time?

Few. A small team with one person in every project cannot run twenty at once, and does not try to. If we are full, we say when the next slot is.

Start with a 30-minute call. Keep what it finds.

You tell us where the hours go. We tell you which two or three jobs AI should take first, in a short written note. Free, in English or French.

ben@bss.mc · a reply within one working day