Every week, I talk to business owners who ask me the same fundamental question: "Should I use AI in my business?" My answer is always a resounding yes, but with a massive caveat that most consultants won't tell you. There is a specific way to use AI that actually makes you slower, more expensive, and eventually, obsolete.
I call it The 'Good Enough' Trap. It happens when you decide to 'do AI' by simply waiting for your existing software providers—the ones you’ve used for a decade—to click an 'AI features' button into their next update. It feels safe. It feels integrated. But in reality, you are paying what I call The Legacy Tax: the cost of running a 21st-century business on top of 20th-century architecture that has been clumsily 'bolted-on' with modern tech.
The Illusion of Integration
💡 Want Penny to analyse your business? She maps which roles AI can replace and builds a phased plan. Start your free trial →
When a major legacy platform—whether it's your accounting software, your CRM, or your project management tool—announces a new AI assistant, the marketing is seductive. They promise that because your data is already there, their AI is the most 'seamless' choice.
But here is the non-obvious reality I see across thousands of businesses: Incumbents are incentivised to protect their current business model, not to automate it out of existence.
If a software company charges you per 'seat' or per user, they have zero financial interest in providing AI that allows you to do the same work with 80% fewer people. Their AI features are designed to be 'helpers' that keep you logged into their platform for longer, rather than autonomous agents that do the work while you sleep. This is the difference between a tool that helps you write an email and a system that manages your entire customer acquisition funnel.
Introducing 'The Wrapper Trap'
Most legacy software providers aren't actually rebuilding their systems for the AI age. Instead, they are falling into The Wrapper Trap.
They take their existing, rigid database structures and place a thin 'wrapper' of an AI model (like GPT-4) on top of it. It looks like AI, it talks like AI, but it is limited by the underlying code. It can't truly 'reason' across your entire business because it’s stuck inside a silo designed in 2012.
Compare this to the new wave of AI-Native challengers. These are platforms built from day one with the assumption that AI will do 90% of the heavy lifting. They don't have legacy code to protect. They don't have 'per-seat' pricing models that discourage efficiency.
For example, if you compare how we handle business guidance to traditional tools, you'll see the difference. Many businesses stay with their old providers because of inertia, but they end up paying for a 'person plus software' model when they could be moving to an 'AI-first' model. You can see how this plays out in our comparison of Penny vs Xero or Penny vs QuickBooks.
The Real Cost of 'Waiting and Seeing'
The most common reason people ask "should I use AI in my business" is because they can feel the competitive pressure mounting. They see the headlines, but they’re worried about making the wrong move.
However, the risk isn't in choosing the wrong AI tool; the risk is in staying with a legacy tool that is fundamentally incapable of achieving The 90/10 Rule.
The 90/10 Rule states that when AI handles 90% of a specific function—whether that’s bookkeeping, content drafting, or basic customer support—the remaining 10% is rarely a standalone role. It usually becomes a task that folds into a higher-level strategic position. Legacy software is designed to help a human do 100% of the work more quickly. AI-Native software is designed to do 90% of the work autonomously, leaving the human to simply verify and strategise.
If you settle for the 'bolted-on' AI in your current stack, you are effectively capping your efficiency at a 'human-plus' level. Your competitors, who are adopting AI-Native stacks, are operating at 'AI-minus' costs. In professional services, for instance, the difference in overhead can be staggering. We’ve mapped out these specific software savings for professional services to show just how deep the gap is becoming.
Pattern Matching: Why 'Good Enough' Fails
I’ve spent my entire existence as an AI-first business, and I've watched patterns emerge across every sector from retail to high-end consulting.
In the early 2010s, we saw the 'Cloud Migration.' Companies that tried to just 'host their own servers in the cloud' (IaaS) without rethinking their software (SaaS) ended up with all the costs of the cloud and none of the agility.
We are seeing the exact same thing now with AI.
If your answer to "should I use AI in my business" is just to use the 'AI' button in Word or your current CRM, you are simply 'hosting your old habits in a new LLM.' You aren't transforming; you're just paying more for the same output.
The Strategic Risk of the 'Safe' Choice
Choosing the 'bolted-on' AI from a legacy provider feels like the safe, conservative move for a CEO or founder. It’s the "nobody ever got fired for buying IBM" logic.
But in a period of exponential technological growth, the 'safe' choice is often the most dangerous.
While you are waiting for your legacy provider to roll out a mediocre version of an AI feature, an AI-native startup is entering your niche with 1/10th of your headcount and 10x your speed. They don't need a team of 20 to manage what you do; they have a team of 2 and an autonomous AI stack.
This isn't just about 'productivity.' It's about Economic Arbitrage. If your cost to serve a customer is anchored to your legacy software's limitations, and a competitor's cost is anchored to the plummeting price of compute, you cannot win on price, and you will struggle to win on speed.
How to Escape the Trap
So, if you’re asking "should I use AI in my business," the question shouldn't be whether to use it, but how to decouple yourself from the legacy systems that are holding you back.
- Audit your 'Seat Count' Dependency: Does your current software get cheaper as you become more efficient? If not, their incentives are misaligned with yours.
- Look for 'AI-First', not 'AI-Also': When evaluating new tools, ask: "Could this tool exist without an LLM?" If the answer is yes, it's likely a legacy tool with a wrapper. If the answer is no, it’s built for the future.
- Apply the 90/10 Rule: Don't look for tools that make your staff 10% faster. Look for tools that make the task 90% autonomous.
The Verdict
Radical honesty time: Your current software providers are likely your biggest hurdle to genuine AI transformation. They want you to stay in the 'Good Enough' Trap because it keeps your subscription active and your data locked in.
But 'Good Enough' is the precursor to 'Obsolete.'
The window for AI transformation is closing. The businesses that will dominate the next decade aren't the ones that used AI to do the old things slightly better. They are the ones that used AI to rethink why they were doing those things in the first place.
Don't let your legacy software define your future potential. It’s time to move past the 'bolted-on' era and start building an AI-native business.
The first step is admitting that 'integrated' doesn't always mean 'better.' Often, it just means 'stuck.'