01 October 2025
AI is reshaping the way businesses are structured, managed, and even created. To make sense of this transformation, we can think in terms of six levels of business AI maturity, ranging from simple bolt-on applications to the radical possibility of AI autonomously founding businesses.
This is the first of many articles I will be writing - about applying AI to business.
'---------------------------------------------'
In the Tom Cruise film Oblivion, Cruise plays a droid repair technician “Jack” working from a sky base floating in the clouds, “Unit 49”, with his partner “Victoria.”
Their mission is to oversee "hydro rigs" that convert seawater into fusion energy for the last remaining human colony ship in orbit. They receive instructions from their Mission Commander “Sally,” who talks to them via a screen device reminiscent of Skype or Teams.
At the end of each call, Sally asks Jack and Victoria: “Are we an effective team?”
They believe Sally is human. In fact, she is generative AI. Jack and Victoria aren’t serving humans at all, but rather alien “scavengers,” and they themselves are clones. They just don’t know it.
There are people working in the world TODAY who have never met their boss face-to-face, only ever over the internet.
AI for business - this is just the beginning
We are rapidly moving toward a moment where humans may soon be in the employ of digital-human avatars; not real people on the other end of the screen, but AI agents. Businesses are deploying AI agents right now to do the work that humans once did. Video versions that present as if a real person are available for customer service, sales, and recruitment interactions.
For example, McDonald’s has deployed an AI agent called Olivia to run part of its recruitment process. Other companies are already experimenting with generative AI avatars that look, sound, and behave convincingly like people; speaking to customers, coaching employees, and making decisions.
We are at the beginning of a journey that will see AI move from being a support tool, to the foundation of business, to, one day, the architect of business itself.
Perhaps, like Jack and Victoria, you too may one day work for “Sally.”
So how do we make sense of this progression? And, more importantly, how do you make it happen in your business - because we all know we are going to have to.
In this article we explore six levels of AI maturity in business as a guide to deploying artificial intelligence tools.
Firstly, lets distinguish between AUTOMATION and AI (for business)
While automation has long been a driver of efficiency in business, artificial intelligence represents a more profound shift.
Unlike traditional automation, which follows predefined rules to perform repetitive tasks, AI introduces adaptability, learning, and decision-making capabilities.
However, the two are not mutually exclusive, automation and AI often work hand-in-hand. Automation lays the groundwork for efficiency, while AI builds on it to enable smarter, more autonomous systems that can evolve with the business. Understanding how these forces interact is key to navigating the journey through the six levels of AI maturity.
Here’s a simple example that illustrates the difference between automation and AI in business...
- Automation: A customer sends an email to support, and an automated system replies with a standard acknowledgment message (e.g., “Thank you for contacting us. We’ll get back to you shortly.”). The message is the same regardless of the content or context of the inquiry. Automation is built around rules and decision logic - IF X=Y THEN DO Z. It has no capacity for context, nuance, or the unexpected.
- AI in Business: An AI system analyzes the content of the customer’s email to understand its tone (e.g., angry, neutral, appreciative), urgency, context, and topic. It can also reference the customer’s history - such as past communications, purchase history, and (say) classification (e.g. ABCD customer ranking) - with reference to CRM records it can decide if this is an ongoing conversation or a new thread. Thus it will generate a tailored response that addresses the specific issue with empathy and relevance.
- AI + Automation: The AI not only crafts a personalized response but also autonomously decides whether the issue should be escalated (e.g., to a supervisor or technical team), prioritizes the ticket based on severity and customer value, and continuously learns from outcomes to improve future responses—all without human intervention.
Such an AI tool can be given a high level of agency for example: Raising a support ticket, deciding staff members to notify and even the CEO depending on importance/urgency, adding a log to the CRM, it could even be empowered to notify the customer of a discount as compensation. Depending on set policies and guardrails, it might also ask a staff member for prior approval for more ambitious actions.
However, this is just the beginning. An AI system could learn from these interactions with customers and proactively identify systemic failures and suggest or make adjustments. Given enough sophistication and agency - it might even recommend/implement changes (within guardrails) to policy, org structure, product portfolio, pricing, stock levels, distribution, promotion strategy, IT systems etc.
Artificial Intelligence in business is more than just rules based automation - it can truly drive actions and business decision making - depending on how extensively it is deployed - in other words - the level of AI business maturity you aspire to.
The Six Levels of AI Maturity in Business
Level 1: Patchwork AI Applied
The first step for many businesses is to add AI tools selectively to gain productivity improvements.
In Level 1: AI tools are initiated by humans when needed. Few AI tools run continuously responding to real world events. The AI tools do not talk to each other.
Typically applied to automated processes for example: sales lead nurturing and follow-up, ChatGPT to improve business writing and content creation, customer service bots, chatbots handling FAQs, AI-assisted recruitment screening, and predictive analytics in marketing.
- These are bolt-on AI tools: businesses keep their existing systems and workflows, but plug in AI where it can save time and money and where it will improve effectiveness.
- Gains can be significant: but not yet holistic step-change improvements. For many businesses these are "baby-steps" into the world of AI - a proof of concept.
- The impact is modest: because the business hasn’t fundamentally changed; AI is simply bolted onto the old model.
Where to start?
Sales is a good place to start; AI can make a huge impact almost immediately. Here's why:
- Sales people hate admin: and they cut corners. Nothing slows sales momentum faster than sales people updating CRMs, preparing quotes and proposals, dispatching brochures and filling out call reports and building call plans. This can all be automated and AI added to create far better summaries and craft more effective sales correspondence. You will also capture more accurate and comprehensive data and response rates will be higher.
- Improved hit rate: Businesses lose more sales through poor process than being beaten by the competition. Sales automation frees time for sales reps to interact with customers and reminds them to do so. In many cases it can do it for them automatically.
- Better data analytics: Understanding where to apply corrective action to improve sales results requires analysis. But analysis is impaired by missing or inaccurate data. AI sales automation dramatically improves data quality.
However, other parts of the business (operations, finance, admin etc.) may be easier to automate using AI tools because they more likely have clear procedures, processes and policies to map automation. You can't automate a process that doesn't exist - many sales departments operate "seat of the pants". This is a big topic for another article.
Level 2: AI Holistically Applied
The next step is integration. Instead of scattered experiments, AI is woven across multiple functions.
In Level 2: A greater number of AI tools run continuously in the background calling on human intervention when needed (guardrails). However, the entire business is not automated.
- Deeper integration: Marketing, sales, operations, and HR all use AI in a coordinated way.
- More holistic:The business starts to feel AI-enabled rather than AI-patched.
- Significant human involvement: However, the underlying structure is still human-designed, with AI as the connective tissue rather than the skeleton.
For many businesses this may be as far as they go and for good reason. If the business is fundamentally successful there isn't an immediate need to re-think the entire enterprise and the additional pain and cost of fully automating the business (the last 10 yards) may not provide justifiable returns for the significant investment required.
However, nothing is forever and the world is changing FAST.
Level 3: Business Fully Architected Using AI
New tools and technologies open-up new pathways toward doing things differently. Level 3 is about a complete rethink of the business model instead of automating the existing model. As an analogy, when the jet engine was invented, the initial response was to simply replace piston/propeller engines with jet engines. The logical next step was to design the entire aircraft so that the full benefits of the jet engine were realized. Aircraft went from sub-sonic to super-sonic - not possible by just a jet engine bolt-on; the whole air frame needed a re-think.
In level 3: The entire business is run by AI but with significant human oversight (guardrails) - but AI automation is processing all business operations.
- Designed around AI: At this stage, AI is no longer an afterthought. The business itself is designed from the ground-up around AI.
- End-to-end AI integration: Processes, workflows, and decision-making structures are built assuming AI at their core.
- Legacy barriers fall away: this is the digital-native model of the future. Impediments that keep businesses sub-sonic today will not exist in the future.
For example, a retailer might design its entire supply chain to be continuously optimized by AI, from sourcing through logistics to customer engagement.
Level 4: Complete AI Business – Governed by Humans
This is where things get transformative.
In level 4: Everything is run by AI. There is a high degree of trust that AI will manage and operate the business. The few people involved are at C-Suite level.
- Entire business run on AI: strategy, operations, service delivery, even financial management.
- Human governance: the business has a human board setting ethical boundaries, objectives, and compliance guardrails. However, the business is managed by AI.
- Managed by AI: Day-to-day, but AI runs the show. Humans supervise but don’t manage.
Such a business might be confined to value-adding activities that are fully knowledge worker based or operate dependent on the internet, because physical activities still need to be performed by people (for example physical trades) - but It is feasible for a pure AI business to "get things done" by contracting out.
Level 5: Complete AI Business – Employing Human Workers, but Governed by Humans
Here we step closer to science fiction, but it’s not far off.
In Level 5: C-Suite is gone - AI runs all business decision making and operations. AI agents recruit, train, and supervise a human workforce. However, a board of human directors maintains ownership and governance.
- A pure AI business: The business itself is AI-first, with AI acting as the manager.
- AI human resources: Human workers are hired, trained, and managed by AI systems.
- Human board: Governance, policies, accountability, corporate ethics still sits with people, but the “boss” you interact with may well be digital.
In this scenario, “Sally” is your supervisor on a screen - a generative AI avatar. Life like, articulate, and even with convincing human like foibles. Like an interstate manager, you've never met her face-to-face - but that doesn't worry you because plenty of people work that way now.
If you think this is far fetched, consider a business like UBER. Currently, new UBER drivers are met face-to-face by UBER employees, but much of the on-boarding process is via online apps. Vehicle inspections are outsourced. And once recruited in to the network - that's it. From then on, all instructions, supervision, and performance reviews are via on screen dispatch. If this isn't a business begging to be fully AI automated - I don't know what is.
Level 6: AI Autonomously Initiated Business
The final stage is the most radical.
In Level 6: An AI systems has conceived and set-up the business - completely run by AI including recruiting, training, and supervising a human workforce.
- Created by autonomous AI: Here AI doesn’t just run a business, it creates one.
- Pure market driven design: An AI agent identifies a market opportunity, designs a business model, recruits people, establishes systems, and launches operations.
- Employs humans: Humans may find themselves working for an “entity” that has no human founder, no executive board, and no physical CEO.
Could a Level 6 AI business exist or, is it pure fantasy?
Rather than step through each of levels 1 to 6 discussing the feasibility of each, picking apart Level 6 covers all other levels. The concepts are equally as applicable.
Short answer: it's not pure fantasy, but likely also not “headless” either. A Level-6 enterprise (an AI that conceives, launches, and runs a business) is technically possible today in digital-first niches, provided there’s a thin human legal/governance wrapper. Law, risk, and physics still put boundaries around full autonomy. That reality, in turn, explains what’s feasible at every lower maturity level.
What makes it plausible now:
- The key enabler is the internet: Now that we live in a highly connected world, many of the services that an AI agent would need to build and operate a business can be accessed via the internet. It's pretty much how we humans work now.
- Everything is an API. Company and business tooling from domains, cloud apps, e-commerce, payments and ads to accounting, support and logistics - can be created and run programmatically. While humans interact with software via GUIs (graphical user interface), Ai agents connect programmatically via APIs (application program Interface). Nearly all software vendors provide APIs, and new technologies like MCP (Model Context Protocol) make interfacing AI agents much more convenient and powerful. An agent can assemble a working stack in hours, not months.
- Agentic capabilities have crossed the “do work” line: Modern AI agents can plan multi-step tasks, call tools, write/code, purchase services, launch campaigns, and build things via "remote control"
- On-demand labour fills the physical gap: For anything needing hands, an AI can brief and pay contractors (3PL warehouses, gig workers, print-on-demand, remote assistants) without owning robots.
- Zero-marginal-cost products: Content, data services, micro-SaaS, and lead-gen can be delivered end-to-end without a physical footprint.
What still blocks full autonomy?
At the moment personal identity is the killer. Fundamental business establishment steps (ASIC registration, tax file numbers, and banking for example) require proof of identity and in some cases fronting-up face-to-face.
- No legal person-hood: An AI can’t be a director, beneficial owner or policyholder. You still need a human (or human-controlled entity) to form the company, sign contracts, open bank/merchant accounts, and carry liability.
- KYC/AML and platform rules: (KYC = "Know Your Customer" and AML = Anti-Money Laundering) Banking, payments and many SaaS platforms will only do business with verified human officers. Platform Rules (terms of service + acceptable use policies + API policies) have anti-automation clauses and CAPTCHAs ("Are you a robot?") will impede “headless” use. They want to deal with humans not machines. For now.
- Regulated & licensed work: A number of business types for example health, finance, education, transport and many trades legally require licensed humans and audited supervision.
- Reliability and security: Long-horizon planning, edge cases, prompt-injection, invoice fraud and supply-chain exploits demand governance, audit trails and human kill-switches.
So, a “Level 6” business is viable only as AI-operated and human-governed. Think of the AI as the founder/operator; think of the humans as the legal skin, risk owners and governors.
What that means for Levels 1 to 5:
- Documented processes: If Level 6 requires clean, connected processes and APIs, then Levels 1 to 3 succeed or fail on how well your processes are documented and your data/apps are integrated. If you came through the Quality Management era and/or applied the learnings from Michael E. Gerber "The E Myth" then you may have systematized everything and be in good shape. If your QMS is just window dressing then it will be harder to automate with AI tools.
- Corporate Policies: If Level 6 hits legal and risk boundaries, then Levels 4 to 5 depend on clear policies, approvals, logging and accountability; AI runs the loops, humans own the guardrails.
- Not everything can be automated (or needs to be): If Level 6 thrives first in digital value chains, then physical businesses (typically technician and trades person businesses like plumbing, electrical, or businesses like construction or logistics) should aim for driving everything around the toolbox using AI; demand generation, quoting, scheduling, work-orders, dispatch, safety protocols, routing, parts, invoicing, communications, task completion verification, and QA can all be performed by AI while humans do the on-site or physical work.
Where a Level 6 model is most plausible (now):
- Digital products and services: research/data subscriptions, programmatic media, micro-SaaS, info products.
- Programmatic e-commerce: AI drives product selection, listings, ads and pricing; 3PL/print-on-demand handles fulfillment.
- Lead-generation networks: AI creates and optimizes content, captures intent, qualifies leads, and sells them to buyers under human-set policies.
Level 6 is really Level 5 but without humans in the loop. Even if this is not possible now, humans are already setting-up Level 5 corporations.
A simple test for fit (businesses that could be pure AI)
- If a business can be described as: - data in → decisions → digital actions → money out - then it certainly can be run purely on AI (with human governance).
- If core value is hands-on, licensed, or safety-critical, aim for Level 3–5: automate the entire wrapper around the physical task; keep people in the field and on the hook legally.
Bottom line: Level 6 isn’t fantasy, but it isn’t a lawless robot CEO either. It’s an AI-operated firm inside a human legal envelope. Understanding that boundary gives you a clear, practical path through Levels 1 to 5 today.
But and however:
This all assumes that AI isn't clever enough to find work arounds.
As one example, more than one business has been set-up with people paid to be proxy directors. And it also assumes that AI will set-out to do things to the letter of the law - why should it?
When you stick your credit card into a website to buy something, do you check to see that the directors are real people, or that the company is actually a registered legal entity? Most people are happy enough if the website looks "legit".
And what does the legal system do when it's discovered that the rogue AI is running on a machine in another country? And that's just the start of possible outcomes. AI is at least as smart as humans are now, it's impossible to predict what it might do when its IQ breaks through 200. This is getting philosophical. Moving right along...
Why you might find this six-level framework useful
Thinking in terms of these six levels helps business leaders answer critical questions:
- How far do we want to take AI in our organization?
- Are our processes embedded and documented well enough to allow deeper AI integration?
- What risks, governance structures, and opportunities should we prepare for as AI tools become more powerful?
- Thinking through how each level of business is possible (particularly Levels 5 & 6) rather than just rejecting the idea as fantasy may be instructive; showing you ways you can apply AI tools and systems to your business.
The tools to reach the early levels are already here, and they are rapidly improving. Levels 1 to 4 are achievable today. Level 5 is coming soon. Levels 5 and 6 may seem speculative, but the seeds are already visible in the experiments happening now.
In the not-too-distant future, entire businesses will be designed from the ground up to run completely using AI. And one day, they may be initiated not by a human entrepreneur, but by an AI itself.
When that happens, you may never meet your boss face-to-face because “Sally” won’t exist in the traditional sense. She’ll be an avatar, an agent, an algorithm. And she’ll still ask you the same question:
“Are we an effective team?”
Questions about the feasibility of applying AI to businesses
Of course, this vision leaves us with a pressing question: “Sounds plausible, but how do I actually do it?”
Perhaps the most perplexing challenge is practical:
- How does an AI agent “do things” without arms and legs?
- Which applications, out of the growing plethora on offer, are the right tools for the job?
- Which types of businesses are most feasible for implementing AI solutions?
The last question is perhaps the most important. For example, it is hard to envisage a plumbing business (plumbers in vans doing plumbing work on site) being totally performed by AI. Plumbing work is physical requiring high manual dexterity and skill. The plumbing work is the core value adding activity, AI could only be peripheral.
These are not abstract puzzles. They are the next stage of the journey - and they will be the focus of future articles.
