Augmentation of Intelligence
Your people operate a layer up, building their own automations and owning the calls AI shouldn't make.
They build their own automations
A session is two hours at the desk, not in a classroom. Each person arrives with one task they do every week (the Monday report, the quote that eats an afternoon, the follow-up nobody sends), opens the tools they already have, and builds the agent that does the repetitive part. We sit beside them. When the two hours are up, something is running on their own machine, with their own data, and they built it. Next week they bring the next task.
- They build their own automations instead of waiting for someone to build it for them
- One real task per session, in their own tools, running before they leave
They know which decisions stay human
Every agent has a line. Below it the agent acts; above it a person signs. Drawing that line is part of the session and it gets written into the agent, so it holds when we're not in the room. The agent drafts the reply; the discount goes through a person. The agent flags the overdue invoice; stopping a shipment is a call. A team that can draw its own line can be trusted with tools we never taught them.
- They know which decisions AI can take, and which ones stay human on purpose
- The line lives in the agent, not in a policy nobody reads
They understand what they built
Each agent ends with a one-page card the person writes themselves: what it does, what it reads, where the line is, what to do when it breaks. That card is the ownership. Before session one we agree the baseline with you, which tasks and how many hours a week per area, and we measure the same thing the same way after handover. So the result is the same number measured twice, and a person who can rebuild, change or switch off what they own.
- They understand what they're building, so it survives without us
- Baseline agreed before session one, measured again after handover
The program
Cohorts run by area, so the examples in the room are always theirs. Each area names a champion who owns what the area builds after we leave. Sessions are weekly, two hours, on the week's real work; between sessions the team runs what it built and brings back what broke. What stays with you: the guides, the templates, and the record of every agent built with its owner. This is what makes a company AI-first. Not the tools, but the people who know how to use their own judgment on top of them.
Phases, not dates.
The order is fixed. The calendar is set once the diagnosis closes.
- Phase 01
Baseline
With each area lead we pick the tasks in scope and measure how long they take today. The number is written down and agreed before anyone opens a tool.
- Phase 02
Cohort sessions
Weekly, by area, two hours, on their own work. Every person leaves every session with something running.
- Phase 03
Their own automations
Between sessions they run what they built, bring back what broke, and start the next one. The champion keeps the area's record.
- Phase 04
Measurement and handover
The same tasks measured the same way. Guides, templates and the record of what was built stay with you, under your champions.
Little, and early.
A short list, all before the first session. None of it takes a day.
- 01A champion per area: the person who'll own what the area builds after we leave
- 02Two hours a week per participant, blocked on the calendar, not squeezed between meetings
- 03Their real tasks and their real tools, with access working before session one
- 04A baseline agreed before session one: which tasks, how many hours, measured how
- 05A leader who shows up, at least to the first session and the last
What people ask about this.
No. A course teaches the tool and sends people back to their desk. Here the desk is the classroom: each session is two hours on a task the person already does, and it ends with that task running on their machine. Nobody takes notes on prompting. They leave with an agent they built.
The people who do the repetitive work of each area, plus their lead. A cohort is one area at a time (sales, finance, operations, people, support), so every example in the room belongs to someone in it. Cohort size is agreed with you. The champion of each area attends everything; the lead comes to the first and the last session.
The ones you already pay for. If the company has Claude, ChatGPT or Copilot licences, we build on those. If it has none, we say which one fits before session one and you open the accounts in your name. We don't host anything for you, so what they build keeps running after we leave.
Hours per week on the tasks in scope, measured before session one and again after handover, with the same method both times. Plus the record: how many agents are running, who owns each one, and which ones are still running a month after we left.
Ready to become AI-first?
One discovery call. We tell you where to start, what it takes and what it's worth. If the answer is no, we say so.