What exists today, 22 September 2026. The Ley de Scale-ups ("Scale-ups Act") doesn't exist yet. The Plan IA360, presented on 21 September 2026, announces its regulatory measures for «otoño de 2026» ("autumn 2026"), with no published text and no exact date within that window. What follows doesn't depend on that law arriving, or on when.
A business turning over twice what it did two years ago doesn't have the same automation problem as one that's just starting out. In the early stage, almost any manual process is manageable because volume is low. The problem appears when that same process, which worked fine by hand, starts eating into the hours of a workforce that hasn't grown at the same pace as orders, invoices or customer enquiries. That's the moment automating stops being a nice-to-have and becomes the condition for continuing to grow without hiring at the same pace as volume.
What the plan announces for growing businesses
Flagship project 9 of the Plan IA360 announces a future Ley de Scale-ups that would continue the Ley de Startups ("Start-ups Act"): it would extend support for the phase after a business's first five years of life by up to five more years, would extend the current seven-year limit for science-based (deeptech) companies, and would expand support and incentives to turn science into business, «como las deducciones por I+D+i y las bonificaciones por contratar investigadores» ("such as R&D&I deductions and incentives for hiring researchers"), without the plan setting any figure. The stated goal is to go from 13 to 50 unicorns by 2030, the United Kingdom's current level, Europe's leader. It's an announced reform, not a bill already going through the legislative process, and it serves here only as context: a growing business needs, sooner or later, to solve the same operational problem that automating solves, with or without a law alongside it.
The signs it's time to automate, not the date on the calendar
There's no company size or turnover figure that marks the exact point. What there is, instead, is a set of signs that, combined, indicate a specific process has reached its manual limit:
| Sign | What it indicates |
|---|---|
| The process's monthly volume grows faster than the workforce handling it | The cost of continuing by hand keeps rising unchecked |
| The process follows clear rules, not case-by-case expert judgement | It's a candidate for simple automation, not a complex project |
| The data lives in documents or conversations, not in a structured system | An extraction step is needed before automating the rest |
| A one-off error is correctable without serious harm | It lends itself to automation with light rather than full human review |
| The process touches several systems that today are connected by hand (copying and pasting between CRM, ERP, email) | It's the most common pattern for cost-effective automation |
The Plan IA360 itself applies this same logic to its business voucher: its roll-out method provides for «una primera fase que permita validar el modelo antes de su escalado» ("a first phase that allows the model to be validated before it's scaled up"), instead of rolling out all at once. It's the same principle worth applying to any internal automation: testing one specific process before extending the solution to ten other similar processes, even though neither exercise depends on the other.
Two steps, not one leap
Not every automation calls for the same thing. When the process follows fixed rules and connects tools that already exist — an email that arrives, a piece of data to extract, a record to create in the ERP — the solution is an automation flow built with n8n and an LLM: low entry cost, data kept on your own infrastructure, and a rollout of a few weeks. For many SMEs, it's the first real entry point into applied AI.
When the process requires chaining several steps with judgement — deciding, not just executing; calling more than one system; drafting a reply that varies case by case — you need the second step: an agent that reasons over each case and leaves the critical action signed off by a person until it demonstrates enough stability. Returns, reconciliation, invoice reading or first-line support are common examples of processes that start with simple automation and, if volume justifies it, scale up towards an agent.
Working out which of the two steps fits each process in your business, before investing in either, is what the what-to-automate-first test is for: you describe up to three processes and get a priority order with an indicative type of solution for each one.
What doesn't depend on any law
None of the above requires the Ley de Scale-ups to come into existence, or to do so with the figures that today are only announced. The moment a process stops holding up by hand depends on your business's volume, not on the regulatory calendar. If that moment has already arrived for a specific process, waiting for the law to be presented doesn't change that arithmetic.
Frequently asked questions
Do I need to be a scale-up, or meet some legal requirement, to automate?
No. Automating processes doesn't depend on any legal category, or on the future Ley de Scale-ups existing. It depends on volume and the type of process.
How do I know whether I need a simple flow or an agent?
It depends on whether the process follows fixed rules (a flow) or requires case-by-case judgement (an agent). The prioritisation test guides that decision by describing the real process, not with a general rule.
How long before you see results?
A simple automation flow is usually rolled out in a few weeks; an agent starts with a pilot of several weeks before moving to production, with a prior evaluation.
Does the Ley de Scale-ups change this decision in any way?
Nothing changes yet: the plan announces regulatory measures for autumn 2026, with no published text. Once it exists, it could affect the funding of these projects, not whether it makes sense to automate them.
Flagship project 9, Development and technology area, of the Plan IA360 (official La Moncloa PDF). For the rest of the plan, see the complete guide to the Plan IA360.
Content last verified: 22 September 2026.