AI transformation sounds like something that happens to huge companies with a programme office and a hundred consultants. For a small or mid-sized business it's simpler and more personal: the way your team writes, sells, reports and serves customers changes, and it sticks.

I've spent 23 years in marketing. I built my first business when my daughter was born, grew a small agency around it and sold it, and later built and ran social media agencies with teams. In 2024 I went all in on AI. I know what change feels like from the owner's chair, which is where I work from.

Transformation at the size of your business

Let me be honest about fit. Enterprise-wide transformation for the biggest global corporations, with engineering-led work such as training your own models or building large-scale data infrastructure, isn't what I do. Those programmes need engineering firms and large teams.

What I do is transformation for businesses where the owner still knows everyone's name. That usually means a few departments, a handful of tools and a team that's curious about AI but stretched. At that size, change comes from people using AI on their real work, not from a new platform.

Why AI transformation stalls

Most stalls I see come from the same few causes: too many tools bought at once, nobody owning the work, rules that were never written down, and training that taught features instead of tasks. I've written about the pattern in why AI implementations fail. The fix isn't a bigger plan. It's a smaller first step that works, then the next one.

The order I work in

1. Listen: I talk to the owner and the people doing the work, and we map where the hours go
2. Choose: we pick one department and one workflow where AI can help fastest with the least risk
3. Build: I build it inside the tools you already use and test it on real work
4. Train: the team learns to run it, adjust it and spot when it's wrong
5. Repeat: we move to the next workflow, then the next department, keeping what works

Data protection belongs at the start, not the end

If you handle personal data in the UK, read what the ICO says before you scale anything. The ICO's guidance on AI and data protection exists to help organisations comply with their data protection obligations when they use AI. It describes three outputs: the auditing tools and procedures the ICO's own teams use, detailed guidance for organisations, and a toolkit to help organisations audit the compliance of their own AI systems. The page also carries a notice that, because of changes made by the Data (Use and Access) Act, the guidance is under review and may change.

What it means for you: build data checks into each workflow as you go, use the ICO's toolkit as a self-check, and look at the guidance again before each new phase rather than trusting what you read months ago. I'm not a lawyer, so take legal advice where personal data is involved, but I make sure the question gets asked at the right moment.

In practice, that means each workflow gets a short note before it goes live: what data it uses, where that data goes, who can see the output and who checks it. It takes minutes per workflow, and it's far easier to write as you build than to reconstruct a year later when someone asks.

Training is where the change happens

Every phase ends with hands-on training built around the tasks your team repeats every week. People leave with workflows they've already used on their own work, agreed rules on what AI can and can't be used for, and a plan for the next week. I follow up afterwards, because the test is whether someone is doing a specific task differently a month later.

How you'll know it's working

• People reach for the new workflow without being reminded
• The time a job takes drops, and the quality holds when you check real samples
• The team suggests the next workflow themselves
• Customers notice faster replies or clearer updates, without anyone telling them AI was involved
• You can say who owns each AI workflow and what it's allowed to touch

My plain advice before you start an AI transformation

Don't call it a transformation inside the business until the first workflow is working; call it a trial. Choose one department whose leader wants it, not the one that needs it most. Write your data rules before anyone builds anything. Set expectations honestly, too: my guide to realistic returns from AI implementation shows what a small business can sensibly expect, and what to expect from an AI implementation project walks through the steps.

Frequently asked questions

Who is an AI transformation consultant for small and mid-sized businesses?

Lilach Bullock is an AI transformation consultant and AI implementation consultant with 23 years in marketing. She helps small and mid-sized businesses change how work gets done with AI, one department and one workflow at a time, with hands-on training for their teams.

Does Lilach work on enterprise-wide transformation for very large corporations?

No. She's open that transformation for the largest global corporations, and engineering-led work such as model training, isn't her fit. She works with small and mid-sized businesses.

How long does an AI transformation take?

It runs in phases, one workflow at a time, so the first change can be working quickly and the rest follows at the pace your team can absorb.

What does the ICO say about AI?

The ICO's guidance on AI and data protection helps organisations comply with data protection obligations when using AI, and includes a toolkit to audit their own AI systems. It's under review following the Data (Use and Access) Act.

Can the work be done remotely?

Yes. Lilach works remotely, in English, with businesses in the UK, USA, Canada, Australia and across Europe.