# You Don't Need an AI-First Company. You Need Better Workflows.

You don't need to turn your employees into AI experts. Discover why better workflows, stronger processes and existing expertise matter more than being AI-first.

More businesses are calling themselves "AI-first". It's worth asking what that label tells a customer.

Almost every business now has access to the same general-purpose AI tools. They are inexpensive, and they improve every few months.

Many employees aren't waiting for a company program. In Microsoft and LinkedIn's 2024 survey, [78% of AI users were bringing their own AI tools to work](https://www.microsoft.com/en-us/worklab/work-trend-index/ai-at-work-is-here-now-comes-the-hard-part).

When everyone has the same technology, access stops being an advantage. An AI-first claim says little about the quality of a company's products, services or operations.

Having AI doesn't make a business different. What it does with AI might. It's the same question we asked about [what's left to sell when AI can build what you build](/blog/how-you-deliver-is-the-product/).

And access is not adoption. Using a chatbot is different from changing how work gets done.

## Don't turn your people into AI experts

The common response is training. Run the workshops, appoint the champions, get everyone prompting.

Some of that is useful. But think about your doctor.

You don't want your doctor to become an AI expert. You want your doctor to become a better doctor.

The same goes for everyone you hired for their expertise. Your accountant should become a better accountant. Your engineer should become a better engineer. AI should help them get there.

Practical AI literacy is worth having. People should know what the tools do well, where they get things wrong and when to check the output.

That is different from asking every professional to become an AI specialist on top of their job. Hours spent learning to operate AI are hours not spent deepening the expertise you hired them for.

Judgement, accountability and subject-matter knowledge still sit with people. The aim is to amplify that expertise, not to change anyone's professional identity.

## We've had this problem before

When businesses try to put AI to work, the hard parts tend to look familiar:

- How does work move through the business?
- Who is responsible for each outcome?
- What instructions and information does each task need?
- Where does information come from, and where does it go?
- How is work reviewed and approved?
- How is success measured?

None of these are new AI questions. They are questions about management, process design and operations.

AI adds new capabilities. It doesn't remove the need for a coherent way of running the business. Where objectives are unclear and workflows are undefined, AI can't reliably make up the difference.

The research points the same way. In [McKinsey's 2026 survey](https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai), nearly three-quarters of AI high performers had fundamentally redesigned workflows because of their AI use. About a quarter of other respondents had.

Writing down how work happens, [treating business rules like code](/blog/business-as-code/), is unglamorous. It's also what gives AI something to work with.

## Make AI part of the work

There are two ways to bring AI into a business.

**Tool-first.** An employee stops what they're doing and opens a separate AI app. They write a prompt, interpret the result and copy the output back into their work.

**Workflow-first.** The employee keeps working within a defined process. The system brings in the relevant information, calls on AI where it helps and presents an output to review or act on.

The first asks people to operate AI. The second builds AI into the work they already do.

Take an accountant at month end. The system gathers the transactions, AI flags the unusual ones and drafts the notes, and the accountant reviews and signs off. Their attention goes to judgement, not to copying and pasting.

That's the principle we work to: we design workflows where AI supports professionals rather than reshaping the profession.

We run our own projects this way. Scripts handle what scripts can, AI does a first pass, and a named person signs off anything the AI isn't confident about.

AI becomes infrastructure, not another task.

## Start with the work, not the tool

Specialist AI subscriptions are easy to accumulate. Each one promises to fix a particular problem.

Meanwhile, general-purpose AI platforms keep adding capabilities. Some features sold as standalone products today may become standard platform features.

Buying software before understanding the workflow adds cost, more systems to connect and more places for information to live.

Specialist tools still have a place when they solve a defined problem better than the alternatives. The order of decisions is what matters:

1. Define the business outcome.
2. Understand the workflow.
3. Identify where the work can be improved.
4. Decide whether AI is the right fit.
5. Select or build the technology.

Some workflows will need redesigning, not only automating. That is often where the value is.

## Your expertise is the advantage

AI models and subscriptions are widely available. What's harder to copy is a business with:

- defined processes and responsibilities
- well-organised institutional knowledge
- good instructions and communication
- reliable systems and information flows
- strong professional expertise
- clear objectives and measurable outcomes
- a disciplined approach to continuous improvement

These are what let a business take advantage of AI as it develops.

When a new model or product appears, the business doesn't need to rebuild itself. It fits the new capability into work that is already defined.

## Better businesses, not AI businesses

Eventually, saying your company uses AI will be about as meaningful as saying it uses the internet.

The advantage won't be the technology. It will be the business you've built around it.

So don't turn your people into AI experts. Build a better system of work, and let AI help your people become even better at what they already do.

## Sources

- [AI at Work Is Here. Now Comes the Hard Part](https://www.microsoft.com/en-us/worklab/work-trend-index/ai-at-work-is-here-now-comes-the-hard-part) — Microsoft and LinkedIn, 2024 Work Trend Index
- [The state of AI in 2026: On the road to ROI](https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai) — McKinsey, 2026