AI in business automation: 5 levels that show where your organisation stands
Ask ten directors whether their organisation uses AI and nine will raise their hands. Ask those same people whether AI has measurable impact on revenue, margins or growth targets and the room goes silent.
Research by McKinsey shows that 88% of companies deploy AI, but only 6% actually measure impact on business results. The remaining 94% experiment, test and apply AI to small tasks without tangible results at the business level.
Raymond Muilwijk, our Group AI Director, observes: "AI is being used as a personal tool, but not as a system that initiates the work itself."
)
)
Technology is not the bottleneck
We have all the models and tools at our disposal. And yet, the majority of AI applications run on an individual's laptop. A workflow is started and stopped by a human. That is precisely the biggest limiting factor of AI right now. As long as people press the start button, you make people faster, but you don't change the system. Does the assistant wait for a prompt? Or does the system pick up signals itself and act autonomously? Only in the latter case does the value of your business automation scale. It's not about how much AI is present in your organisation. It's about how that AI works.
)
Five levels of AI adoption
There are five levels of AI adoption help you understand where your organisation stands.
Tools
Your organisation uses AI daily in applications: CRM, email, content tools. Your team works faster, but the process stays the same.Data
Within your organisation, AI is connected to business data. One clear answer replaces seven screens. This is where it starts to become useful, and where it goes wrong if your data isn’t right.Trigger
You implement workflows that start with a signal, not a person. The system acts: signal → enrichment → decision → action → learning. This is the tipping point.Redesign
The process is reinvented or disappears entirely. New touchpoints emerge, assets generate themselves, and it's the system, not the editor, that initiates the work.Agentic
Agents act proactively without you directing them: prospecting leads, following up, qualifying, all on their own initiative.
In practice, we see that many organisations find themselves somewhere between levels 1 and 2. That's the point at which you need to start thinking about how to progress.
)
Level 3: Trigger
The tipping point is at level 3: Trigger. At this level, the system initiates its own work for the first time, instead of a human.
An example of removing the human from the critical process loop is the Enexis case. A grid operator delivering targeted campaigns at exactly the right moment, without anyone starting the process. The system triggers itself based on weather data, peaks in energy usage and a news agent adding real-time context. The result: +19% engagement and an AMMA Award.
Level 4: Redesign
Level 4 is where it gets truly interesting. It's no longer about automating what you already did. It's about rethinking the process itself. Process steps you once took for granted suddenly turn out to be redundant, because the system no longer needs them. You work towards this by first truly mastering levels 2 and 3. Redesign the process without having data integration and automated workflows in order, and you'll find the foundations are missing to keep the system running reliably.
)
Level 5: Agentic
This is what most of the hype is about. At this level, autonomous agents prospect, qualify, follow up and close transactions without human intervention. The possibilities are real, but there's a flip side.
Agents are non-deterministic: the same input doesn't always produce the same output. And errors in agentic systems compound exponentially. A small mistake in your data gets magnified thousands of times by an agent before anyone notices.
At iO, we see agentic systems as R&D territory for now. You can already experiment with them, particularly in areas like software engineering or strictly defined tasks. But if you want to move seriously into level 5 tomorrow, you'd be wise to get levels 2 and 3 right today, to the point of tedium.
From maker to director
The human side of this shift deserves attention, because automation always raises the same question: what does this mean for our people? At each level, your relationship with AI changes:
Levels 1 and 2: this tool makes you faster
Levels 3 and 4: the machine does the work, you make sure it's right
Level 5: you set the framework, the rest takes care of itself
The role shifts from maker to director. You go from executing every step yourself to determining how the system works, what boundaries it has and whether the output reflects the intent. People don't become less valuable. They deliver a different kind of value.
)
Take the step towards AI as an autonomous business system
The organisations making a difference with AI at the business level have one thing in common: they've put the system to work instead of the people. The technology is there. But the step from AI as an individual tool to AI as an autonomous business system requires a conscious choice:
First, get your data in order
Then set up triggers
Then think about which processes can be done differently
We guide organisations in taking that step, with systems that actually run and grow.
Curious about what level your organisation is at, and what the next step looks like? We'd be happy to talk it through with you. Get in touch and we'll schedule a conversation.