Most business owners who write to us start the same way: "I know AI can help me, but I'm not technical and I don't know where to start." And the answer is always the same: it is easier than you imagine, and the first step has nothing to do with technology.
In 2026, AI tools for business work like a spreadsheet or a CRM: you do not need to understand how they are built to use them and get real results. What you do need is clarity about what you want to automate and a willingness to spend a few hours on the initial set-up.
This guide is for anyone who wants to do it themselves: from identifying what to automate to having the first workflow running with no-code tools, with concrete examples and the tools most used by Spanish SMEs right now. If what you are after is how AI is rolled out in a company with a team that builds it to measure, read how to roll out AI in your company: the 5 phases.
The biggest myth about AI: "this is only for big companies with a technical department"
This myth has less and less basis in reality. According to INE, in the first quarter of 2025, 13.4% of Spanish companies with fewer than ten employees were already using some artificial intelligence technology, 5.9 points more than the year before. Those companies do not have a technical department. Five years ago the myth was true: implementing AI required data engineers, your own servers and budgets that only a large company could afford. Today, the same capabilities are available through visual platforms that anyone with basic computer skills can learn to use in days, not months.
The confusion comes from mixing up two different things: building AI and using AI. Building artificial intelligence models does require advanced technical knowledge. But using those models to automate your business is as accessible as using WhatsApp Business or setting up a Shopify shop.
What has changed is that the layers of technical complexity are now hidden behind visual interfaces. You define the business logic (what has to happen, when and under what conditions) and the platform takes care of the rest. If you can explain a process out loud, you can automate it.
AI does not require you to know how to code, just as using WhatsApp does not require knowing telecommunications. Today's platforms work with visual logic: "when X happens, do Y". The learning curve is much shorter than you imagine.
The real obstacle is not technical: it is conceptual. The businesses that move fastest are those that spend time identifying which problems they want to solve before touching any tool. Those that fail start the other way round: they look for tools without knowing what they want to automate.
The 3 types of AI any business can use today
Not all AI is the same. So as not to get lost in the market noise, it helps to understand what kind of problem each category solves and which one fits your business best right now.
Conversational AI: chatbots and assistants
These are the systems that understand natural language and hold conversations with your customers or team. The best-known example is ChatGPT, but in a business context we are talking about chatbots trained specifically for your business: they know your products, your prices, your way of working and can answer questions, handle enquiries and qualify leads autonomously.
They are deployed mainly through WhatsApp (the most used channel in Spain for communicating with customers), on the website as live chat, or by email. If in your business the same questions are answered every day (prices, opening hours, availability, order status), this is the type of AI people usually start with. You can explore how our chatbots work before making any decision.
Process automation AI
These are the systems that connect tools and run tasks automatically when an event occurs. Examples: when someone fills in your contact form, the system automatically creates a lead in your CRM, sends a personalised welcome email, creates a task for the salesperson and adds the contact to a follow-up sequence. All without anyone pressing a button.
This type of AI is especially powerful for repetitive workflows you currently do by hand: invoicing, customer follow-up, order management, data synchronisation between tools, internal notifications. If in your business someone spends hours copying information from one place to another, there is a clear opportunity here. AI agents for SMEs take this concept a step further.
Generative AI: text, images and analysis
These are the systems that create new content or analyse information to draw conclusions. In a business context, the most useful applications are: writing emails and sales proposals, generating product descriptions, analysing customer reviews to identify patterns, meeting summaries, and creating social media content.
It is not about letting the AI do all the work: it is about using AI as a copilot that speeds up tasks that today take you hours. A self-employed person who writes every proposal from scratch can start from a draft that already carries their tone, their terms and the client's details, and spend the time on what really decides the sale.
Step by step: your first automation without coding
These are four steps that, in this order, avoid the most expensive mistakes of someone starting on their own, whatever the sector or size of the business.
- Identify your 3 most time-consuming processes. For a week, write down everything you do repetitively: answering the same questions, moving information between tools, sending reminders, doing follow-ups. At the end of the week, rank them by time consumed. The top three on that list are your candidates for automation. Do not start with the most complex or the one that attracts you most: start with the one that steals the most time.
- Pick a single process to start with. The most common mistake is trying to automate everything at once. The first project has to be small enough to finish, and representative enough for the team to see the value of AI and lose their fear of change. A basic customer service chatbot, an automatic welcome flow for new customers, or an appointment reminder system: any of these is a good first project.
- Choose the right tool for that process. Not all tools are good for everything. In the next section we explain the ones most used by Spanish SMEs and what each is best for. The general rule: do not choose the most sophisticated tool, choose the one that gets you something working in the least time. You can improve it later.
- Measure the impact and then scale. Before switching on any automation, define how you will measure whether it works. If it is a chatbot: how many enquiries does it answer without human intervention? If it is a follow-up flow: how many qualified leads does it generate? If it is automation of administrative tasks: how many hours a week do you save? With that data in hand, in the second month you have clear arguments to decide what to automate next and, if you work with an agency, to negotiate better.
This process seems obvious, but most businesses skip it. The result is AI projects that never reach production because they became too complex before proving their value. Start small, measure, and then expand.
The no-code tools most used by Spanish SMEs in 2026
The ecosystem of no-code AI tools has matured a lot. These are the ones we use most often with our clients and the reasons we choose them.
Make (formerly Integromat)
What it is for: connecting tools and automating complex workflows without code. Make works with visual "scenarios" where you drag and connect blocks. When something happens in one application (an email arrives, someone fills in a form, a sale is created), Make can trigger actions in dozens of other tools.
Ideal for: businesses that use several tools (CRM, email marketing, spreadsheets, WhatsApp Business) and want them all to "talk" to each other. It has integrations with more than 1,500 applications. The free plan covers basic use cases; beyond that you pay a subscription, and what marks the jump in plan is volume: how many operations your scenarios run per month and how often they fire.
Learning curve: moderate. In 2 to 3 hours you can have a first scenario running. Make has an active community in Spanish and there are tutorials for practically any use case.
n8n
What it is for: the same as Make but with more technical flexibility and a self-hosting option. n8n is especially popular among businesses that handle sensitive data (health, legal, finance) and prefer their data not to pass through third-party servers.
Ideal for: businesses with more complex workflows, teams with at least one person with some basic technical knowledge, or companies with strict data privacy requirements. The cloud version of n8n is paid; the self-hosted version is free but requires managing a server.
ChatGPT API through intermediate platforms
You do not need to connect directly to the OpenAI API (which would require some technical knowledge). Platforms such as Botpress, Voiceflow or Stack AI let you set up advanced AI assistants with visual interfaces, using OpenAI or Anthropic models underneath without the user having to worry about the technical side.
These platforms are ideal for creating chatbots trained on your business information (product catalogue, frequently asked questions, returns policy, prices) that respond coherently and usefully, not like a generic decision-tree chatbot. If you want a deeper look at what they can do, read our article on how the self-employed are saving time with AI in 2026.
WhatsApp Business API
For many businesses in Spain, WhatsApp is the main channel for communicating with customers. The WhatsApp Business API (through certified providers such as Twilio, 360dialog or Vonage) lets you connect WhatsApp to automation systems and AI chatbots.
Important: the official WhatsApp Business API is not the free app you can download on your phone. It is a version for businesses that allows advanced automation, multiple agents and connection with other tools. The activation process requires verification by Meta and usually takes between 3 and 10 working days. Once active, it is the foundation on which most customer service chatbots for Spanish SMEs are built.
The 5 mistakes when setting up your first no-code automation
In the first projects someone sets up on their own, the same patterns of error repeat. Knowing them in advance saves you months of frustration.
- Mistake 1: automating a process that should be simplified first. If a manual process is chaotic, automating it only makes it more chaotic and faster. Before automating, simplify. If the appointment management process has 12 steps and none is documented, the solution is not a chatbot: it is first to document and simplify the process, and then automate it. The rule: if you cannot explain the process to someone in 5 minutes, it is not ready to be automated yet.
- Mistake 2: choosing the most powerful tool instead of the most suitable one. You do not need the most advanced tool on the market. You need the one that solves your problem reliably and that you can maintain. Many businesses spend weeks learning complex tools when a simpler option would have solved the problem in days. A tool's power is worth nothing if you do not put it into production.
- Mistake 3: not involving the team from the start. Automations that fail usually do so because the team that has to use them did not take part in designing them and sees them as a threat or an imposition. Involve the people affected from the design phase, explain what problem it solves (not what technology it uses) and make someone on the team the "owner" of each automation.
- Mistake 4: not defining success metrics before starting. Without metrics, it is impossible to know whether the automation is working. Define before launching: how many hours do you want to save? How many enquiries should the chatbot resolve without human intervention? How many leads does it need to qualify per month? These metrics are also the basis for justifying the investment and deciding what to automate next.
- Mistake 5: thinking that "switching on" the automation is the end of the work. Automations need maintenance. Chatbots improve over time as they accumulate conversations and you spot patterns of questions you did not know how to answer. Workflows are adjusted when business processes change. Set aside time (or budget) for ongoing maintenance, because an abandoned automation can become a source of errors.
None of these five mistakes is technical. All are about judgement: what you automate, with which tool, with whom and how you measure whether it works. That is why they can be avoided before touching anything. If you are going to do it with a supplier, the failures that cost most are in the 7 mistakes when rolling out AI in a small company.
Do I need an agency or can I do it myself?
The honest answer is: it depends, and both options can be right depending on the context.
You can do it yourself if: you have time to learn (the first project takes up a few evenings), the process you want to automate is relatively simple and well defined, you are comfortable with digital tools in general, and you have tolerance for the initial mistakes that are part of the learning process.
The advantage of doing it yourself is cost and learning. No-code platforms have free or very cheap versions, and the knowledge you gain in the first project speeds up all the following ones. There are active communities in Spanish, YouTube tutorials and documentation in Spanish for most of the main tools.
It makes sense to hire an agency if: your time is worth a lot and you cannot afford to spend weeks learning tools, the process you want to automate affects critical areas of the business (sales, invoicing, customer service) and mistakes have a real cost, you need complex integrations between multiple systems, or you know the project will be left half done between the day-to-day emergencies.
What you should avoid is hiring an agency without understanding the basics. You do not need to be a technical expert, but you do need to understand what you are asking for, how to measure whether it works and who will maintain the automation once delivered. An agency that teaches you nothing about what it is building creates dependency, and dependency is expensive in the long run. On our services page we explain how we work so that the client always understands and controls what we have implemented.
The hybrid option, starting with an agency for the first project and then continuing in-house with the knowledge gained, usually gives the best results in SMEs with small teams.
Frequently asked questions
How much does it cost to implement AI in a small company?
It depends, and what is useful here is not a range: it is knowing what it depends on. There are three items. The tools, which are paid by subscription and rise with the volume you move (operations per month, conversations the chatbot handles, users), and which almost always have a free plan to try. The design and implementation, which is paid once and depends on how many systems have to be connected, how many exceptions the process has and, above all, whether it is already documented or has to be organised first: that last one is what makes a project most expensive and what nobody mentions. And maintenance, monthly, which almost nobody budgets for and everybody ends up needing. To know whether it pays off, do not start with the price: time for a week how much time goes on that process today and put on it the value an hour has in your business. You have that number, not an agency. We discuss prices on a call, once we know what has to be built; any figure before that is an invented number.
How long does it take to see real results?
There is no standard timeline, and be wary of anyone who gives you one before having seen your process. What sets the pace is not the technology: it is initial clarity. A chatbot with a tightly defined scope (the questions you already know by heart and a protocol for what it cannot answer) is among the first things that can be up and running. A workflow that touches several systems, one of them undocumented, goes much more slowly, and the time goes on organising the process, not on building it. Projects that start with "I want to automate something" are the ones that never finish.
What happens if the chatbot makes a mistake or gives wrong information?
Well-configured chatbots have escalation mechanisms: when they cannot answer something with certainty, they do not make it up; they transfer the conversation to a human or ask the customer to make direct contact. The key is in the initial configuration: clearly defining what the chatbot can and cannot answer, and setting a clear protocol for situations outside its scope. Mistakes happen, as they do with people; what matters is that the system detects and handles them gracefully.
Do I need to change all my processes at once?
No. In fact, it is exactly the opposite of what we recommend. The most successful AI implementation is incremental: one process at a time, well measured, before moving on to the next. Changing everything at once creates chaos, resistance from the team and difficulty identifying what works and what does not. The digital transformation of a small business is built process by process, not in one go.
Which process should I automate first?
The quick answer: the one that steals the most time and is the most repetitive. The most common candidates in Spanish SMEs are replying to frequent enquiries on WhatsApp, following up leads who have not replied, sending appointment or payment reminders, and generating recurring reports. If you are not sure which to choose, do this exercise: for three days, every time you do something more than once, note it down. At the end you will have your prioritised list.
Xora IntelligenceAI consultancy and custom software in Los Barrios (Cádiz). We write about what we build for businesses in the province. If you have a question about your case, write to us.




