What I Actually Believe About AI in Business

I get asked this question more than any other: "What do you actually think is real about AI in business, and what is hype?"

It's a good question. I spend every day building AI systems for service businesses — which means I see both sides clearly. I've seen AI change companies in ways that are genuinely remarkable. I've also seen AI deployments that made things worse, not better.

Here's what I actually believe.

What AI Is Genuinely Good At (In Business)

Speed and consistency at routine, rule-based tasks.

AI is extraordinarily good at tasks that are routine, repeatable, and well-defined. Responding to a lead enquiry within 90 seconds. Sending an appointment reminder at the right time. Logging a call to a CRM. Generating a draft proposal from a template. These tasks don't require creativity or judgment — they require speed and consistency. AI delivers both at scale.

The human cost of these tasks isn't just time — it's cognitive load. Every time a team member has to remember to send a follow-up, log a contact, or book a reminder, they're spending mental energy that could go elsewhere. AI removes the cognitive overhead of routine operations.

24/7 availability.

Humans need sleep. They need weekends. They get sick. They have off days. AI doesn't. For any business where inbound enquiries arrive outside office hours — which is every business — AI availability is a genuine competitive advantage.

The lead that comes in at 9:30 PM Friday is just as valuable as the one that comes in at 10:30 AM Monday. AI treats them the same way. Humans, structurally, cannot.

Handling volume without proportional cost.

A human salesperson can handle a finite number of leads. An AI system can handle 1,000 leads with the same marginal cost as 100. For businesses with seasonal or unpredictable volume spikes, this is enormous.

What AI Is Genuinely Bad At (Right Now)

Anything requiring real relationship or trust.

Complex sales involving significant spend, nuanced client situations, emotional conversations, trust-building over time — these require humans. An AI can conduct a qualification conversation. It cannot build a relationship.

The businesses that try to automate the parts of their business that fundamentally require human connection either succeed in the short term and damage their brand, or fail quickly.

Genuine strategic judgment.

AI can surface data, summarise options, and identify patterns. It cannot make genuinely strategic decisions — the ones that require understanding of human motivation, competitive dynamics, timing, and risk tolerance in combination.

Treating AI output as a strategic recommendation without human review is how businesses make expensive mistakes. AI is a very fast analyst. It is not a strategist.

Creative breakthroughs.

AI can generate creative content. It cannot originate genuinely new ideas. The most compelling strategies, positioning breakthroughs, and product innovations still come from humans thinking deeply in context. AI can help develop and refine, but it rarely originates.

What I Worry About

I worry about businesses deploying AI to cut human interaction as a cost-saving measure rather than to augment human capacity.

There's a version of this technology that makes businesses colder, more transactional, and less human — and that version makes businesses worse. It reduces the relationship quality that professional service businesses compete on. It signals to clients that they're not worth a person's time.

This version exists. Some businesses are building it. The short-term cost savings are real. The medium-term damage to client relationships and brand is also real.

The right frame for AI in professional services: it should make the humans in your business more effective at the human parts of their work, not less needed for the human parts of their work.

An AI front desk that handles intake and scheduling means your senior consultants spend their time on deep client work, not admin. That's the right version. An AI system that handles deep client relationships too — that's the wrong version.

Why I Started Irtiqa

I started Irtiqa because I kept seeing a specific problem: genuinely excellent service businesses losing revenue not because their work was bad, but because their operational infrastructure was broken.

Great clinicians losing patients because nobody answered the phone after 5:30. Excellent consultants losing deals because the follow-up email never got written. Superb agencies losing clients because the onboarding process was chaotic.

These businesses deserved better infrastructure. Not to replace what made them good — their expertise, their relationships, their craft — but to protect and amplify it by fixing the operational gaps that were bleeding revenue before it reached them.

That's the version of AI I believe in. Not AI as a replacement for human excellence. AI as the infrastructure that lets human excellence show up reliably, at scale, without burning out the people delivering it.

What's Coming That I'm Watching

Agentic systems getting better at complex reasoning. The gap between what AI can handle today and what it will handle in 18-24 months is significant. Tasks that require human judgment today — not because they're complex in principle, but because AI isn't good enough yet — will be automated within 2-3 years.

Voice AI reaching true naturalness. Voice AI today is good. In 18 months it will be indistinguishable from human in most contexts. This will transform phone-based business interactions more than anything since the smartphone.

AI in regulated industries normalising. Healthcare, legal, financial services are moving cautiously but they are moving. The compliance frameworks are being built. The deployment patterns are being established. In 3-5 years, AI-assisted intake, triage, and follow-up will be standard in all three sectors.

My Honest Advice

Don't fear it. Don't over-hype it. Deploy it strategically.

Find the parts of your business that are routine, volume-intensive, and don't require human relationship or judgment. Automate those. Free up your humans for the work that genuinely requires them. Measure the impact. Iterate.

Do it in the right order. Capture layer first. Conversion layer second. Operations layer third. Lifecycle layer fourth. Build the foundation before the sophistication.

Be transparent with clients and prospects. Honesty about AI involvement builds trust. Deception destroys it.

And remember what makes your business good: the expertise, the judgment, the relationships, the care. AI is infrastructure. Infrastructure serves the work. The work is still done by people who care.

Book a free audit call if you want to talk through what this means for your specific business — honestly, without a pitch.