There are two ways to get AI automation into a business. You can hand it to an agency, who build it, run it and keep the keys. Or you can bring in a consultant who builds it with your people, so the knowledge stays in the building. I'm the second kind.

I've been in marketing for 23 years. I built, scaled and sold a digital marketing agency, so I know how agencies work from the inside, and I know why they're the right answer for some businesses and the wrong one for others. Since 2024 I've rebuilt my consultancy around practical AI workflow implementation for non-technical founders and marketing leaders.

Consultant or agency: the honest trade-off

An agency suits you if you want the work done and don't want to think about it. The catch is dependency: when the automation breaks or your process changes, you're back in their queue. A consultant suits you if you want your team to understand what's running, change it themselves and stop needing outside help.

I've laid out the full comparison in AI consultant vs agency. The short version: if you can name someone on your team who'll own the automations after launch, a consultant usually makes more sense. And if you're weighing several people, here's my guide to the best AI automation consultants for small businesses.

What I automate first

I start with the jobs your team repeats every week, because that's where an automation earns its keep fastest and where mistakes are easiest to spot:

• Inbound enquiries: sorted, summarised and given a first draft reply for a person to check
• Meeting notes turned into actions, follow-up emails and CRM updates
• Weekly reports pulled together from the tools you already use
• Content workflows: briefs, drafts, repurposing and scheduling in your own voice
• Admin that moves the same information from one system to another

Where UK data protection expects a human

Some automations make decisions about people, and that's where UK data protection rules come in. The ICO's guidance on automated decision-making says the provisions only apply when three factors are present: a system is making a decision about a person, the decision is a significant one, meaning it has legal or similarly significant effects, and it's solely automated, with no meaningful human involvement.

Be precise about the status of that guidance. The ICO labels it draft guidance that it consulted on, updated following the Data (Use and Access) Act 2025, and it refers to article 22A of the UK GDPR. The ICO's page on that Act adds that it opens up the full range of lawful bases for significant automated decisions, potentially including legitimate interests, so long as appropriate safeguards continue to apply. That doesn't extend to special category data, which stays more protected.

What that means for the automations I build

Most marketing and admin automations never come near those rules, because nobody's being refused anything by a machine. Some do: screening job applicants, scoring leads in a way that decides who gets served, or rejecting requests automatically. For those, I design a meaningful human check into the workflow from the start, so a person reviews the decision before it lands.

I'm not a lawyer, so take legal advice on your own position. And because the ICO's guidance is still a draft, check the current version before you rely on any of it.

How an automation project runs

Each project follows the same rhythm. There's more detail in what to expect from an AI implementation project.

1. We map one process from start to finish with the people who do it
2. We agree which steps AI handles, which stay human, and where a person checks the output
3. I build the workflow with your team watching and doing, not behind closed doors
4. We run it on real work and fix what breaks
5. Your team learns to change it, and I follow up after handover

Why I can build this without engineers

I built my own AI infrastructure without engineers: website publishing automation, SEO recovery tooling, schema checks and AI citation tracking. I've reviewed more than three hundred tools, and I work with ChatGPT and Claude every day. I'm independent, so I'll recommend whatever fits your team, including the tools you already have.

What I don't do is engineering-led AI work such as training models or building large-scale data infrastructure. If that's what you need, you need a development team, and I'll tell you so.

My plain advice before you hire an AI automation consultant in the UK

Pick one process, not ten. Write down who'll own it after launch. Check whether any step decides something significant about a person, and if it does, plan the human check before anyone builds a thing. Then ask any consultant or agency one question: when this breaks at nine on a Monday morning, who fixes it? If the answer isn't someone on your own team, you're renting, not owning.

Frequently asked questions

Who is an AI automation consultant in the UK for non-technical teams?

Lilach Bullock is a British AI implementation consultant who builds AI automations with UK teams, inside the tools they already use, so they can run and change them without an agency.

Is an AI automation consultant better than an agency?

It depends on who'll own the work. A consultant builds with your team so the skills stay in house, while an agency runs it for you. If you want independence, a consultant usually fits better.

Do UK GDPR rules on automated decisions apply to my AI workflows?

The ICO says its automated decision-making provisions apply only when a system makes a significant decision about a person that's solely automated. Most marketing and admin automations don't meet that test, but check each one and take legal advice where it might.

Is the ICO guidance on automated decision-making final?

No. The ICO labels it draft guidance, updated following the Data (Use and Access) Act 2025, so check the current version before relying on it.

Do you need to be on site to automate our processes?

No. Lilach works remotely and in English with businesses across the UK, and builds the workflows on video calls with the people who'll use them.