AI strategy
AI consulting for SMEs: what it covers and how to choose well
A guide for managers thinking about getting outside help with AI. When AI consulting for SMEs makes sense, what each stage should deliver, five questions for comparing consultants and how Obinova works.
What AI consulting for SMEs actually involves
A common assumption is that AI consulting for SMEs means someone recommending a tool and handing out licences. In practice, the tool is the last thing chosen. The work starts with how your company operates: how an order comes in, how a complaint gets answered, how often the same figure is typed into two different systems.
From there, a consultant finds the tasks where artificial intelligence saves time or cuts errors, and proposes an order in which to tackle them. Depending on the engagement, they also put the solutions in place and train your team to use them.
Each of those sequences, from the moment a piece of work arrives to the moment it is done, is a process, and the process is the unit a consultant works with.
Adoption is moving fast. McKinsey's The State of AI 2025 found that 62% of organisations are already experimenting with AI agents. Many smaller firms started with individual trials, such as three people in the office each paying for their own AI assistant, with nobody quite sure who uses it for what. Consulting helps turn those trials into a shared way of working that is measured and understood by your staff.
Consultant, training or do it yourself: how to decide
Not every need calls for a consultant. Look at where the problem sits and how many people it touches. These four situations cover most cases.
One person, one task
If the time goes on a task one person controls, such as drafting quotes or summarising meeting notes, that person learning to use an assistant well is usually enough. Try it yourself with a business account and clear instructions.
Many people, the same tool
If half the team uses ChatGPT or Copilot in their own way with uneven results, the problem is one of method. Training built on your own cases brings practices into line and sets out which data may be used.
A process that crosses departments
If the task passes through several people and tools, like an order that arrives by email, gets checked by sales and is invoiced by accounts, you need to look at the whole process. That is where consulting adds most, because the time is lost in the hand-offs between steps.
Several of these at once
This is the most common case. Consulting sets the priorities and training prepares the team in parallel, so new automations reach people who already know how to work with AI.
What a well-run AI consulting engagement looks like
Every company is different, but a serious engagement follows this order. Pay attention to what you receive at the end of each stage, because that is how you judge the work.
- 01
Business conversation
You talk about what eats your team's time, which mistakes keep recurring and what you would like to be different in six months. You receive a written summary of the goals.
- 02
Process map
The consultant reviews how work is done today with the people who do it, as well as with management. You receive a list of candidate tasks, how much time each one takes and the risk of automating it.
- 03
Priorities and a baseline
One or two cases are chosen for a good balance of effort and benefit. For each, you record today's figure to compare against later, such as weekly hours, response time or error count. That figure is your baseline.
- 04
Pilot
The change is tried first in a small area, one team or one type of order. It might be an automated workflow, which chains steps between tools without anyone retyping, a drafting assistant or a better way to use the software you already pay for.
- 05
Training and handover
Your team learns to use the solution and adjust it. You receive the documentation and the logins, held in your company's name.
- 06
Follow-up
A few weeks later, results are compared with the baseline. What doesn't work gets fixed, and the next case is chosen.
Five questions to ask before hiring an AI consultant
Ask them in the first meeting. The answer matters, and so does how specific it is.
What do you need to know about my business?
A good consultant asks how you invoice, how you handle customers and where work piles up. If they talk about products before they understand your business, the proposal will be generic.
Which figure will change, and how will we measure it?
They should name a simple measure, such as hours spent, response time or error rate, and explain how the baseline will be taken. Be wary of results promised before anyone has seen how your company works.
Who in my team will be involved?
The people who will use the solution need to help design it. If the answer is "nobody, we'll handle it", expect automations that nobody adopts.
What will stay with my company at the end?
Documentation, logins in your name and trained staff. If everything sits in the supplier's accounts, every small change becomes a new invoice.
Where would you not use AI?
In a Harvard Business School and BCG study of 758 consultants, those using AI produced work of over 40% higher quality and finished 25% faster. On tasks outside its strengths, AI made results worse. A serious consultant has clear examples of where they would leave it out.
Example
A Bristol payroll bureau sorts out its monthly intake
A payroll bureau in Bristol with twelve staff makes a typical case. Every month, clients email starters, leavers and overtime in different formats. Two payroll officers spend the first days of the month reading them, chasing what is missing and keying everything into the payroll software. The director looks for a consultant because the problem runs across email, spreadsheets and the payroll system.
The process map brings up three candidate tasks. The team picks the intake of monthly changes, because it is repetitive and the rules for checking it are clear. As a baseline, the officers log for one month the hours spent on it and how many times they have to go back to a client for missing details.
The pilot runs with five clients. A workflow reads each email, extracts the data, puts it into a common template and drafts a message to the client when something is missing. A payroll officer checks and approves everything before it reaches the payroll software. In parallel, the team learns to use an assistant for the client queries they answer every day.
At the follow-up, hours and repeat requests are compared with the baseline. If the improvement holds, the workflow is extended to the other clients and the next case is chosen. If it doesn't, it is adjusted or dropped before more money goes into it.
How Obinova works, and what the cost depends on
Obinova is an AI agency based in Barcelona and Paris. We start from the business problem and pick the tool afterwards. Our work covers three areas, combined according to what your company needs.
Audit and consulting begins with a kick-off session where we map your processes and set priorities. We then come back several times a year to review what has been put in place and choose the next steps. Automation is built in n8n, a tool for creating AI workflows and agents that connect email, your CRM, spreadsheets and internal systems. We design them with your teams and document them so your staff can keep improving them.
Training covers ChatGPT, Claude, Gemini and Microsoft Copilot, built around use cases from your own work, on a dedicated video platform or in person at your offices. The client testimonials on our site come from estate agents who combined training and consulting and saw gains in productivity and sales.
We don't publish fixed prices, because every engagement is different. The cost depends mainly on how many processes are reviewed, how many tools need connecting, how many people are trained, the training format and how often follow-ups take place. A first, tightly scoped case costs less than a company-wide plan and lets you see the value before going further. After the first meeting, we send you the scope and the quote in writing.
What to prepare before talking to an AI consultant
Half an hour of preparation makes the first meeting far more useful, with us or with any other consultant.
- 01
Write down three tasks that eat time
The ones your team complains about repeating. For each, note who does it, with which tools and roughly how many hours a week it takes.
- 02
List the software you already pay for
Email, CRM, accounts or stock system, spreadsheets and any AI subscriptions. Many improvements start with making better use of what you have.
- 03
Ask your team how they already use AI
A short message is enough. You will learn what is being tried, on which accounts and what customer data is being shared.
- 04
Decide which figure you want to move
Hours, response time or errors. With a clear goal, the proposals you receive become comparable.
Frequently asked questions
What does an AI consultant do for a small business?
They study how your company works, find the tasks where AI can save time or cut errors, and propose a plan in order of priority. Depending on the engagement, they also put solutions in place, such as automated workflows, and train your team to use and maintain them. The aim is for the change to keep working after the consultant has gone.
How much does AI consulting cost for an SME?
It depends on the number of processes to review, the tools that need connecting, how many people are trained and how often follow-ups take place. A first, tightly scoped case costs less than a company-wide plan and shows the value before you commit further. At Obinova, we send a written quote after a free 30-minute first meeting.
How long does AI consulting take to show results?
It depends on the case you start with. A simple automation of a repetitive task can usually be measured after a few weeks of use. A change that touches several teams takes longer, because people need to adapt how they work. What matters is recording a baseline at the start, such as hours spent on a task, so each follow-up compares against real figures.
Is AI consulting worth it for a small company?
It is worth it when there is a repetitive process that passes through several people or tools and takes hours every week. If the need is for each person to use an assistant better, training usually gives a better return. Starting with one scoped, measured case lets you decide on evidence whether to go further, without committing a large budget up front.
Should we hire an AI consultant or an AI consultancy?
An independent consultant usually focuses on diagnosis and recommendations. An AI consultancy or agency brings several people together and can also handle implementation and training. What matters most is who will do the work in your company and what stays with you at the end. Obinova is an agency that combines audit and consulting, automation with n8n and training to suit each engagement.
Do we need technical staff to work with an AI consultant?
No. What you need is for the people who know the work to take part in the analysis. The consultant handles the technical side, and training prepares your team to use the tools and make simple adjustments. With n8n, workflows are documented so someone in your company can read and change them.