AI consultants, and how to put AI to work in a business

The title covers very different people, so it pays to ask which kind you are talking to. Some advise: they interview your team, write a roadmap and an AI use policy, and run workshops on tools such as ChatGPT, Claude and Microsoft Copilot. Some build: they connect the systems you already use, such as your CRM, Google Sheets, Slack and the ad platforms, through workflow tools like n8n, Make or Zapier, and add a language model for the steps that need judgement. A few also run what they build, week after week, and answer for the result.

Whichever kind you hire, the work should begin with a real job. When the Office for National Statistics asked UK firms what had held back their use of AI, the reasons named most often were difficulty identifying a use for it, the cost, and a lack of expertise. A consultant earns the fee by removing the first of those: naming the one job where AI will pay, in your own numbers, before any tool is bought.^[1]

Far more people search for an AI consultant than for AI implementation, in both countries: most buyers look for someone to trust before they look for a method.

Starting small is the step most projects skip. A common failure is a licence for everyone and no single job that anyone measures, so a few months later nobody can say what changed. We looked at the common reasons in why AI marketing projects fail early.

Allegiance Industries, a facilities company, had sales reps researching prospects one at a time across LinkedIn, company websites and industry directories. In 2026 we built a prospecting system that runs every morning: the team adds target companies to a Google Sheet, Make passes each one to a Relevance AI agent, the agent decides which job titles to look for in manufacturing or education, pulls the contacts from Apollo, and writes each lead into Zoho CRM after checking it is not already there.^[3]

The job was narrow, which is what made it measurable. Each daily run researches up to 41 companies and fills 17 data points per lead, and the client estimates the system saves $9,000 to $10,000 a year. A company with no valid contact goes to a person for review instead of being dropped, and the only manual step left is adding names to the sheet. The Allegiance Industries case study shows each part.^[3]

The same pattern worked on inbound leads. Before he founded the company, Jakub built a lead processing system for a global technology platform as a freelance consultant: each lead from a Meta form fired a webhook, a Relevance AI agent researched the person and their company, and the record went into the CRM with the same fields every time. Follow-up went from 1 to 3+ days to the same day, across 1,600+ leads in the peak quarter, as the lead automation case study describes.^[4]

Neither build needed custom software. For most small and mid-sized businesses the parts are a workflow tool such as n8n, Make or Zapier, a language model such as Claude or GPT for the step that needs judgement (reading, sorting, summarising or drafting), and the systems already in place. Custom code comes in only when a tool cannot reach the data.

It is also where firms that use AI have gone first. In the US Census Bureau's 2026 survey of how businesses use AI, sales and marketing was the most common place to use it, named by 52% of the firms that use AI in any business function.^[5]

If you have not started, you are not late. Official surveys in both countries find a minority of firms using AI at all, with the largest firms furthest ahead, and most of those that use it have so far adopted only one or two kinds of AI.^[1][6]

Good first jobs in marketing repeat, have a baseline, and can be checked by a person in minutes. These are the ones we look at first:

Leave for later anything that would reach customers without a person seeing it, anything whose data you cannot get at, and anything with no baseline to measure against. Whoever you hire, settle these points before the work starts:

This is also how we work. AI for Marketing runs marketing for ecommerce, SaaS and B2B service businesses in the UK and the US on an AI system Jakub Cambor built: the system does the production, Jakub reviews the work, and the client approves every change before it goes out. We also build custom automations like the ones on this page, as an add-on to that managed service rather than a separate consulting offer. Jakub has worked in marketing since 2018 and is a Certified Relevance AI Partner.^[7]