Chapter 2 of 5
What AI-native means.
In a traditional business, people carry work from one step to the next. In an AI-native business, AI agents carry the routine work, and your people handle the decisions and the unusual cases.
- 1Work arrives, one piece after another
- 2Each piece waits at the gate for a person
- 3A person decides, one piece at a time
Most companies use AI beside the work. Few put it inside.
More than 80 percent of companies say generative AI has made no material contribution to their earnings yet (McKinsey, 2025). The reason is where the AI sits. It writes the meeting summary and drafts the email, so one person is a little faster. The work itself still moves from person to person by hand.
AI beside the work
Where most companies are today
- Summarises the meeting
- Drafts the follow-up email
- Answers a question in a chat window
One person saves a few minutes. The quote, the invoice and the report still wait for the same people.
AI inside the work
Where an AI-native business is
- Prepares every quote from your price rules, for a person to approve
- Matches each invoice and sends only the exceptions to a person
- Builds the Friday report, so a person only reviews it
The work stops waiting. Your people keep the decisions and the unusual cases.
The Agentic Ascent. How much responsibility has each part of your business earned?
That is the one question our maturity model asks, of quotes, invoices, reports and every other part of the work. There are six levels in three stages, and each level holds only once the one below it does. Follow one customer quote up the Ascent: as you scroll, each level changes who carries its five steps.
Stage 1 of 3: Fragmented
Level 0, Disconnected
Systems and work stay separate. Every quote waits for the person who knows the next step.
- Steps that wait for a person
- 5 of 5
- Quotes waiting right now
- 0
One quote, as an example. Not a client result.
Where is your business today? Most are lower than they think.
A company can have AI on every desk and still be at Level 0, because the AI sits beside the work. We start by making sure everyone trusts the numbers, because an agent only makes a wrong number move faster.
Fragmented
AI helps people. The work still waits on them.
Most companies we meet start here
- Level 0
Disconnected
You’re here if answering a customer means opening three systems, and the numbers in each don’t match. AI writes the emails and summarises the meetings, and the work still waits on the same people.
- Level 1
Connected
You’re here if your systems are linked and the data lands in one place, but a person still decides what happens to every piece of work. Connection alone is not adoption.
- Level 0
Trusted foundation
One set of numbers, with owners and controls.
- Level 2
Reconciled
You’re here if “booked”, “won” and “margin” mean the same thing in every report, and each definition has a named owner. Agents can now read, check and draft.
- Level 3
Owned & secured
You’re here if every capability has an owner, access is controlled, every action is logged, and nothing changes live without a person’s approval.
- Level 2
Agentic
Agents carry the routine work, within limits a person sets.
- Level 4
Watched
You’re here if agents watch the work every day and send each exception to the person who owns it, with the evidence attached.
- Level 5
Guardrailed autonomy
You’re here if proven plays run on their own in lanes a person approved. Your people set the objectives and keep authority over every exception and change.
- Level 4
Two ways to run the same business.
Most teams buy AI tools. Few rebuild how the work moves. Scroll, and watch each part of the business change.
- HandoffsWork waits for whoever knows the next stepRoutine work moves on its own
- NumbersEach system shows a different numberEveryone works from the same numbers
- ReportingThe Friday report is rebuilt by handAgents prepare it. A person reviews and approves.
- ExceptionsUnusual cases get lost in inboxesUnusual cases go straight to the person who approves
- AIAI is a tool someone triesAI agents have a set job, clear rules and a person in charge
- GrowthGrowing means hiringYou take on more work without hiring at the same pace
AI-native, in short.
What does AI-native mean?
In an AI-native business, AI agents carry the routine work, and your people handle the decisions and the unusual cases. In a traditional business, people carry work from one step to the next by hand.
How is that different from using AI tools?
Most companies use AI beside the work. It writes the meeting summary and drafts the email, so one person is a little faster, but the work itself still moves from person to person by hand. An AI-native business puts the agents inside the work.
What is the Agentic Ascent?
Our maturity model. It asks one question of quotes, invoices, reports and every other part of the work: how much responsibility has it earned? There are six levels in three stages, and each level holds only once the one below it does.
Can a company with AI on every desk still be at Level 0?
Yes, because the AI sits beside the work. We start by making sure everyone trusts the numbers, because an agent only makes a wrong number move faster.
Who makes the decisions in an AI-native business?
Your people. Agents carry the routine work, inside limits your people approved. Your people set the objectives, judge the exceptions and own every live change.
This is an AI-native business. People doing what only people can do.
Agents carry the routine work, inside limits your people approved. Your people set the objectives, judge the exceptions and own every live change. Find out where you stand: it starts with a free report on how your competitors use AI.
